Documente Academic
Documente Profesional
Documente Cultură
INEFFECTIVE MANAGERS
by
PETER CAMMOCK
-;;:::-
NOVEMBER 1991
Submitted to the University of canterbury for the degree of
PhD.
c. '\-
Dedicated to Liz, Mike, Lucy, Mackenzie and Alice
TABLE OF CONTENTS
CHAPTERS
Abstract
1. Overview
1.1
1.2
1.3
1.4
1.5
Introduction
Why study managerial effectiveness?
Finding a research setting
Research approach
Conclusion
2. Literature Review
2.1
2.2
2.3
2.4
2.5
Introduction
What is a Manager?
What constitutes effective versus ineffective management?
How do the characteristics and behaviours of effective managers
vary across different levels?
Conclusion
3. Repertory Grid Technique
3.1
3.2
3.3
3.4
3.5
Introduction
Background to the repertory grid
Personal construct theory
Application of the repertory .grid technique
Conclusion
4. Data Gathering
4.1 Introduction
4.2 Organisation setting and participants
4.3 Sample definition
4.4 Generalizability of findings
4.5 Interview study
4.6 Questionnaire study
4.7 Measuring effectiveness
4.8 Conclusion and overview of data analysis
Page
1
2
2
5
6
12
13
16
28
46
50
52
54
57
61
70
71
72
74
76
80
89
95
100
5. Defining the Characteristics and Behaviours of Most and Least Effective
Managers
5.1
5.2
5.3
5.4
Introduction .
Data analysis
Results and discussion
Conclusion
6. Factor Analysis of the Nineteen Characteristic and Behavioural Categories
6.1
6.2
6.3
6.4
Introduction
Factor analysis
Results and discussion
Conclusion
102
103
109
131
133
134
139
149
7. Exploring Variations in Effectiveness Dimensions Between Management Levels
7.1
7.2
7.3
7.4
7.5
7.6
Introduction
Hypotheses
Interview data analysis
Questionnaire data analysis
Results and discussion
Conclusion
151
152
154
161
162
175
8. Review of Key Findings and their Implications for Management Development at
the MBA Level .
8.1
8.2
8.3
8.4
Introduction
Review of findings and their relationship to past research
Implications for managerial teaching and development
Conclusion and recommendations for further research
Acknowledgements
Bibliography
178
179
190
201
206
208
Appendices
1. An example of a repertory grid interview transcript
2. Management effectiveness questionnaire
3. The characteristics and behaviours of most and least effective managers
4. Pearson's correlation analysis for the nineteen scale variables
Tables
4.1 Interview and questionnaire respondent sample
4.2 Interview respondents by office, level and sex
4.3 Interview elements
4.4 Level of focus requested of interview respondents
4.5 Questionnaire versions
4.6 Questionnaire response rates by office
4.7 Questionnaire respondents by office, work area, managerial level
and sex
5.1 Effectiveness scale categories (characteristics and behaviours)
5.2 Effectiveness scale mean scores, rankings and standard deviations
5.3 Significance of scale mean score differences for most effective manager ratings
5.4 Significance of scale mean score differences for least effective manager ratings
5.5 Most and least effective scale correlation coefficients
6.1 Matrix of S-Index values for two factor solution (North Island/mail survey and South
Island responses)
6.2 Matrix of S-Index values for two factor solution (most effective and least effective
responses)
6.3 Matrix of S-Index values for three factor solution (North Island/mail survey and South
Island responses)
6.4 Factor loadings most/least effective manager ratings
6.5 . A three factor model of managerial effectiveness
6.6 The two sub-categories of fador one (conceptual ability)
6.7 U nrotated factor matrices for most and least effective manager ratings (two factor
solution)
7.1 Average rank of scales and factors across respondent levels (interview responses)
7.2 Scale and factor mean scores for most effective ratees across respondent levels
7.3 Scale and factor mean scores for least effective ratees across respondent levels
. Figures
2.1 A model of managerial effectiveness
3.1 A completed repertory grid
7.1 Average rank of factors across respondent levels (interview responses)
7.2 Factor mean scores for most effective ratees across respondent levels
7.3 Factor mean scores for least effective ratees across respondent levels
ABSTRACT
Recent reviews of the management literature have expressed
concern over the lack of attention to the issue of
effectiveness, This study addresses this deficiency by
describing the characteristics and behaviours of effective
versus ineffective managers in a large New Zealand public
sector organisation (the Department of Social Welfare) .
Repertory Grid interviews were conducted with 89 respondents
.
in four offices of the organisation. A panel of judges
sorted the constructs into a questionnaire which was
administered to a further 365 respondents, Analysis of the
questionnaire data redu6ed the 170 items into 20 scales
descriptive of the characteristics and behaviour of most and
least effective managers in the Department. Factor analysis
of the scales revealed a three factor structure, suggesting
that effective managers require ability in the conceptual,
interpersonal and technical areas, Both the scales and the
factors demonstrated a high degree of interaction, lending
support to previous research findings that emphasize the
holistic, and interactive nature of managerial work.
Significant variations in emphasis on the scales and factors
were apparent between lower and more senior level
respondents. The thesis concludes by considering the
implications of these findings for management education and
development and recommending avenues for further research.
1
CHAPTER ONE
OVERVIEW
INTRODUCTION
This chapter provides an overview of the thesis. It
outlines why it was undertaken, describes the key steps in
the research and overviews the thesis structure. It is
hoped it will clarify the broad purposes of the thesis and
provide a guide to subsequent reading.
WHY STUDY MANAGEMENT EFFECTIVENESS?
The institution of management has become one of the most
influential social forces of the twentieth century. Most of
the goods and services we consume and the jobs we perform,
fall under the direction of managers. As a consequence,
managers have a crucial impact on our happiness and
wellbeing as individuals, families and societies. New
Zealand, in the 1980's, has witnessed an extraordinary rise
in the power and veneration of managers. This is evident,
for example, in the numerical increase in the numbers of
managerial positions, in the rise of what Jonathan Boston
(1991, p.9) calls, "the managerialist revolution", in the
size of Chief Executive remuneration packages (Loomis 1982,
2
verespej 1989), in the rise and influence of bodies such as
the Business Round Table, and in the growth of Masters of
Business Administration programmes. Through the 1980"s, New
Zealand has also witnessed a substantial increase in
concentrations of power and ownership in a small number of
large New Zealand organisations led by business managers
(Hamilton 1991). Clearly this is an institution worthy of
serious academic study and inquiry.
Curiously, the growth of management as a social institution,
has not been accompanied by a similar growth in our
understanding of managerial work. This is most apparent in
the dearth of research on managerial effectiveness. While
there is a huge research literature on the what and how of
what managers do, there is very little written on what
effective managers do. Most of the reported research makes
no attempt to measure the performance of the managers being
studied, or to test the effectiveness of their activities,
in terms of desired outcomes. Three recent major reviews of
the management literature (Martinko and Gardner 1985, Hales
1986, stewart 1989), express concern about this this
deficiency. All three emphasise the need for more research
specifically on the characteristics and behaviours of
effective managers.
The lack of understanding of managerial effectiveness has
not impeded the growth of the management development
3
industry. The training and development of managers has
become a billion dollar international industry. In New
Zealand demand for consultancy services in this field,
escalated through the 1980's, particularly in the rapidly
changing public sector. In the universities, this same
growth has been echoed 'in the development and growth of the
Master of Business Administration degree (MBA). It is
perhaps symptomatic of our lack of understanding of
effective management, that the MBA has come under intense
criticism over the last twenty years. critics of the MBA,
such as Livingston (1971), Hayes and Abernathy (1980),
Leavitt (1983) and laterly Mintzberg (1989) argue rather
persuasively, that MBA programmes are actually reducing,
rather than augmenting, the effectiveness of u.s managers.
Some of these critics go so far as to suggest, that the MBA
process in the U.S.A, with its strong emphasis on rational
quantitative approaches, is undermining the organisational
competitiveness of that nation. Our New Zealand
universities cannot claim exemption from such criticisms, as
our MBA offerings in many cases replicate those of overseas
programmes.
These issues provided the context of this research. While
the criticisms of the MBA process are the subject of debate,
they created sufficient unease to warrant a response on our
part. In a decade which witnessed a great increase in
demand for management development and education, we did not
Use of the term lweI does not imply mUltiple authorship
4
want to be in a position of offering irrelevant or incorrect
material. As management educators, our concern was to learn
about specifically effective and ineffective management in a
changing New Zealand society. The hope was that we could
then translate that understanding into a more meaningful
teaching offering.
FINDING A RESEARCH SETTING
5
In 1984 the new Labour Government precipitated a pertod of
substantial change for New Zealand organisations. New
Zealand society since that time, has experienced a level of
change perhaps unprecedented in its history. This change
impacted heavily on public sector organisations. These
organisations were shaken from the quiet, secure,
bureaucratised mode of operation, which had characterised
their activity for decades, into an era of restructuring,
redundancy, accountability and performance. It is not
surprising therefore, that we were asked in late 1985, to
assist the Department of Social Welfare in the development
of its managers. This project required a substantial
training needs analysis and provided an ideal opportunity to
pursue our research interest in management effectiveness.
The Department of Social Welfare is a large public sector
organisation which, in 1985, employed around 6000 staff.
The Department has offices allover the country, working in
three main service areas, namely Benefits and Pensions,
social Work and Administration. Our brief, was to explore
the training needs of managers at supervisory, middle and
senior management levels.
RESEARCH APPROACH
Literature Review
The study commenced with a thorough review of the
. literature. Our purpose here was to learn what was already
known about management effectiveness and find a point of
departure for the study. As we have mentioned above, we
found that very little work had actually been done in this
area. This finding confirmed the need for further research.
The literature review is presented in chapter two of this
thesis. out of the literature review, we developed the
following research question as a focus of the study; uWhat
are the characteristics and behaviours of effective versus
ineffective managers and how do these characteristics and
behaviours vary between different management levels?UU
Repertory Grid Technique
In conducting the interviews we used an approach called the
Repertory Grid Technique. This technique provided an
important foundation for the research. We adopted the
technique mainly for its capacity to minimise observer bias.
6
It proved to be an exceptionally powerful interview
technique which revealed a wealth of interesting data. The
Repertory Grid is described in detail in chapter
three.
Data Gathering
We commenced our data gathering with a series of interviews,
with managerial and non-managerial staff, in the Manukau,
Hamilton, Nelson and Christchurch offices of the Department.
The spread of offices was designed to produce a sample which
would be as representative as possible of the total
Department. Most of the interviews were conducted by the
author and took one to two hours to complete. In all, we
conducted 89 interviews, 88 of which proved usable. The
interview respondents were asked to differentiate between
effective and ineffective managers in the Department. Their
responses were recorded in the form of bi-polar constructs,
which differentiated the effective and ineffective managers,
in the interview comparison. The following is an example of
one of these constructs;
Visible; walks the floor
and spends time with
staff.
Seen infrequently
by staff; less
visible.
7
Over three hundred of these constructs were generated in the
interviews. The constructs provided an important data
source in themselves. In addition they were used to form a
questionnaire, which was used to gather additional data from
a larger sample of the Department's staff. A group of six
judges sorted the interview constructs into twenty-one
initial logical categories, which were used in the
questionnaire development process. A final sorting process,
produced a questionnaire with 170 items all comprised of
constructs generated in the Repertory Grid interview
process. The was presented to respondents in
two identical sections, each with. 170 items. On one section
they were requested to rate the least effective manager they
knew at a designated level and on the next, the most
effective manager.
The questionnaires were administered by the author, to staff
in the same four offices in which the interviews had been
conducted. By visiting the offices to distribute the
questionnaires and personally following up on the
respondents, we were able to get a very high level of
response (greater than 80%) from the staff in those offices.
The sample was strenghtened by a mail survey to a further 60
respondents. These respondents were followed up by
telephone, yielding a mail survey response rate of 63%. In
all, usable responses were recieved from 365 staff in both
managerial and non-managerial levels. These responses
8
returned a total of 730 questionnaires, half describing most
effective managers at various levels and half describing
least effective managers. The data gathering phase was
exhaustive and exhausting. It took place over two years
during 1986 and 1987. The data gathering phase of the
r e s e a r c ~ is detailed in chapter foura
Describing the Characteristics and Behaviours of Most and
Least Effective Managers
It was clearly not feasible to separately analyse the
responses to each of the 170 questionnaire items. Some form
of item reduction was therefore necessary. Beginning with
the categories developed previously by the six judges, the
170 questionnaire items were assembled into twenty logical
categories. Using correlation analyses and measures of
inter-item correlation, we reduced. the 170 questionnaire
items into twenty robust scales. These scales provided an
excellent description of the characteristics and behaviours
of both effective and ineffective managers in the
Department. They covered areas such as team building,
consultation, overview, and innovation. Examples of each of
the characteristic and behavioural categories were obtained
from the interview data, to round out the picture. The
scales and the procedure used in developing them are
described in chapter five.
9
Factor Analysis
We used factor analysis to further reduce and model the
twenty scale categories. We identified a three factor
structure which described three broad abilities required to
manage effectively in the Department. These were conceptual
ability, interpersonal ability and technical ability. The
conceptual ability factor fell into two logical sub-
categories, namely intuitive and analytical ability. The
factors and the procedure used in identifying them are
described in chapter six.
Exploring variations Between Management Levels
The scales and the factors were then used to explore
differences in the way effective and ineffective management
was construed by respondents at the non-managerial,
supervisory, middle and senior management levels. Firstly,
we used the interview data, which was content analysed using
the twenty scale descriptors. A count was made of the
number of times each of the scales had been referenced by
non-managerial, supervisory, middle and senior management
respondents. We used this information to compare the
relative emphasis on the scales by interview respondents at
each of the management levels. For the questionnaire data,
we calculated the mean scores given to each of the scales
10
and factors by non-managerial, supervisory, middle and
senior management respondents. This-enabled us to explore
the patterns and significance of variations in emphasis
between the four levels. The procedures used and the
results of this part of the study are detailed in chapter
seven.
T.he Implications of Our Findings for Management Development
The research described in the first seven chapters indicates
that managerial effectiveness (in addition to analytical and
technical ability) is heavily reliant on interpersonal and
intuitive ability. Overall, the findings highlight the
deficiencies of processes that over-emphasize
(as claimed by critics of the MBA) rational, analytic
approaches to management. What emerges is a model of
managerial effectiveness based on interpersonal ability but
requiring an additional balance of intuitive, analytical and
technical abilities. This finding suggests the need for an
MBA process that provides for the development of the
individual's interpersonal and intuitive skills and
insights, as well as catering to more conventional technical
and analytic development. Chapter eight reviews the main
findings of the study and details their implications for
management development at the MBA level.
11
CONCLUSION
Overall, we feel that the study is successful in addressing
the research question, in that it provides a lucid
description of the characteristics and behaviours of
effective versus ineffective managers. It has also been
successful in describing variations in emphasis on those
characteristics and behaviours between management levels.
Its findings have spurred us into ongoing research in the
area and have made a significant contribution to our efforts
as management development professionals. We hope that the
thesis will prove of interest to the reader and that its
findings will make a worthy contribution to an area of
understanding that has been, until recently, largely
neglected by academic researchers.
12
CHAPTER TWO
LITERATURE REVIEW
INTRODUCTION
The growth of management as a social institution is one of
the more notable phenomena of the twentieth century (Burnham
1941, Chandler 1977, Kanter 1977, Kotter 1982, Meyer and
Zucker 1989). In the West individual hero-managers have
come to assume an almost totemistic quality (Meindl, Ehrlich
and Dukerich 1985). In New Zealand, as in most Western
nations, managers are ascribed responsibility (with its
commensurate payment packages) for the material destiny of
large sectors of our society. Chandler's (1977, p.4)
observation that "rarely in the history of the world has an
institution grown to be so important in so short a time",
seems fully justified.
In the face of the growing veneration of the management
institution there is evidence that the influence of
individual managers has been overstated (Gamson and Scotch
1964, Eitzen and Yetman 1972, Lieberson and O'Connor 1972,
Pfeffer 1977, Salancik and Pfeffer 1977, Pfeffer and
Salancik 1978, Meindl et al 1985, Williams, Chapman, Findlay
and Tuggle 1990). A related tendency has been to abstract
managers from the social, environmental and institutional
13
constraints that limit their influence and impact on their
behaviour (Willmott 1984, stewart 1982, Barnes and Kriger
1986, Hosking and Fineman 1990, Martinko and Gardner 1990).
While there are clear constraints on the potential
contributions of managers there is evidence that management
does matter, and that it accounts for a significant amount
of the variance in organisation outcomes (House and Baetz
1979, Weiner and Mahoney 1981, Smith, Carson and Alexander
1984, High and Achilles 1986). These findings are supported
by most of the studies that attempt to ground managerial
behaviour in empirically determined measures of performance
(see for example Burgoyne and Stuart 1976, Kotter 1982,
Luthans, Rosenkrantz and Hennessey 1985).
These studies indicate "that managerial behaviour is related
to effective org.anizational performance" (Martinko and
Gardner 1990, p.331). Consequently some individual managers
will have a more positive impact on their organisations than
others. From this we may advance the proposition, outlined
by Martinko and Gardner (1990, p.331) that "there are
differences in the behaviours of highly effective and less
effective managers". This is a common sense proposition
that underpins most management theory (Lewin and Minton
1986) .
14
Surprisingly, the managerial research literature provides
little understanding of the behaviours and/or
characteristics of specifically effective or ineffective
managers. We have a very sUbstantial literature on
managerial practise but very little that attempts to
distinguish between effective and ineffective practise. The
definition of specifically effective versus ineffective
management is one of the most important and most neglected
areas of managerial research (smith et al 1984, Martinko and
Gardner 1985, Hales 1986, stewart 1989).
This thesis seeks to make a contribution in this area. It
addresses the following research "What are the
characteristics and behaviours of effective versus
ineffective managers and how do these characteristics and
behaviours vary between different managerial levels.
UU
The term characteristics refers to the personal qualities
and traits of the manager. These include intelligence,
aptitudes, knowledge, values, temperament and personality
characteristics. The term behaviour refers to the way
managers conduct themselves in their observed actions
towards others and in their responses to various job
situations. The inclusion of managerial characteristics in
the research question acknowledges the fact that there is
more to managerial work than just observable managerial
behaviour (Hales 1986).
15
The issues explored in this chapter relate to the research
question. We first address the question "What is a
manager?" by defining the term manager and describing the
defining characteristics of managerial work. Secondly we
ask the question "What constitutes effective versus
ineffective management?" We discuss current literature in
this area and then look at more specific definitional and
measurement issues. Finally we address the question "How do
the characteristics and behaviours of effective managers
vary across different managerial levels?"
WHAT IS A MANAGER?
of the Term "Manager"
The Oxford English Dictionary (1989, p.294) defines a
manager as "one who manages ... a person, or one of a bodY,of
persons, responsible for the general working of a public
institution". This definition acknowledges the specific
responsibility held by the manager "for the operation of a
discrete organisational unit" (Hales 1986, p.109). It
implies that the manager will be vested with formal
authority to run the unit and will in turn be held
accountable to some higher level authority.
These concepts are embodied in the definition of Stewart
(1976, p.4) that a manager is "anyone above a certain level,
16
roughly above foreman whether ... in control of staff or not".
Such a nominalist definition provides a realistic starting
point for research and has been w ~ d e l y adopted by management
researchers (Hales 1986). It has obvious limitations
however, given the "diversity in the composition of
managerial work" (Hales 1986, p.l07).
The most serious limitation of this definition is its
failure to recognise the manager's need to achieve "results
through other people" (Heller 1972, in stewart 1986, p.ll).
As implied in the dictionary definition the manager has
responsibility, authority and accountability to do some
things him/herself. The manager is further distinguished in
that "being assigned more work than he can do, [he] is
authorized to get some of that work done by others for whose
work he is in turn accountable" (Jaques 1976, p.64).
In summary we may define a manager as a person, usually
titled manager, who has responsibility, authority and
accountability for a discrete group of people charged with
achieving a specific set of tasks and objectives. This
embodies distinct responsibilities which the manager has to
fulfil directly. It also implies authority and the need to
get other people to do things for which the manager remains
finally accountable. As noted by Hales (1986, p.ll0) this
implies "a crucial distinction, within the generic term
Imanagerial work', between what managers themselves do and
17
what managers have to ensure others do". Both are essential
components of any definition of the term manager.
"What Do Managers Do?18 Features of the content and Process
of Managerial Work
Further definition can be given to the term manager by
examining what managers do in fulfilling their
responsibilities. The work of the manager appears to have
specific content and process features by which it may be
distinguished from other social functions. We will employ
Whitley's (1985, p.344,345) definition, which describes
content as lithe common behaviours managers engage in as they
carry out their job ... responsibilities" and process as the
"characteristics (e.g., such as duration of activity, mode
of communications, mode of contacts) found among managerial
jobs". For example, planning, budgeting and selling are
content features. The hours worked and the patterns of
communication used in pursuing those activities are process
features.
There is only "moderate agreement" between findings on work
content (Whitely 1985, p.344). Hales (1986, p.93) in a
major review of the literature refers to a level of
"discontinuity, even inconsistency" in these findings. The
research findings on work process are more consistent, with
the core findings remaining constant across "studies
18
conducted in different countries and at different time
periods" (Whitely 1985, p.345). The following features are
representative of the key findings in this area (see Hales
1986 and stewart 1989 for recent reviews) .
Content Features
1. The content (and process) of managerial work varies
across management levels, job types, organisations,
environments and cultures (Burns 1957, Dubin and Spray 1964,
Horne and 1965, Nath 1968, Child and Ellis 1973,
Mintzberg 1973, Boyatzis 1982, Stewart 1982, 1988, Pavett
and Lau 1983). Work content and process can also vary
significantly between different managers performing
jobs (Stewart 1976,1988, Stewart, smith, Blake and
Wingate 1980).
2. There is substantial choice available to the manager in
both what is done and how (Stewart 1976, 1982, 1988, Stewart
et al 1980). The observational studies of Dalton (1959) and
Sayles (1964) for example, both report attempts by managers
to enhance their jobs by altering their content. In many
instances the level of choice is such that a key part of the
manager's work lies in defining the meaning of their
particular job (Gowler and Legge 1983).
19 .
3. Managerial jobs require technical/specialist and more
general managerial skills (Hales 1986). Managers need both
technical understanding and the ability to balance
technical/specialist with more generalist managerial roles
(Kotter 1982, Dakin and Hamilton 1986).
4. Despite criticism (Braybrooke 1964, Mintzberg 1990) the
classical functions (see Gulick 1937, Fayol 1949) still have
validity in terms of the tasks they describe. More recent
research tends to confirm that managers do plan, organise,
command, coordinate and control (see Carroll and Gillen 1987
for a review of this issue). The classical writers however,
imply a work p r o c e s ~ which is unrealistic. Managers in
practise plan, organise, command, coordinate and control in
ways vastly different to those implied by the classical
writers (Kotter 1982).
A number of the later content listings can be seen as a more
dynamic conceptualisation of the classical functions. As
noted by Hales (1986, p.95) Kotter's task listing, along
with those of Sayles (1964) Mintzberg (1973) and Stewart
(1976, 1986, 1988) "all provide fresh insights and
subtleties to the tasks of 'planning', I co-ordinating , and
lcommanding'".
5. Managerial work has a strong informal and political
dimension not accounted for by the classical writers (Dalton
20
1959, Fletcher 1973, stewart 1983, Luthans et al 1985, Hales
1986, Hosking and Fineman 1990). Managers spend a
sUbstantial amount of their time "accounting for and
explaining what they do" in informal and political
interactions (Hales 1986, p.104).
6. More recent research places a greater emphasis on the
development of external contacts, than was the case with the
classical prescriptions (see for example Hemphill 1959,
Mahoney, Jerdee and Carroll 1965, Tornow and Pinto 1976,
Dakin, Hamilton, Cammock and Gimpl 1984).
7. The last fifteen years has seen an increasing emphasis
on change and innovation. The literature is a
good example of this developing emphasis (see for example
Zaleznik 1977, 1989, Adair 1983, Bennis and Nanus 1985, Bass
1985, 1988, Kouzes and Posner 1987, Kotter 1988, 1990,
Bennis 1989).
8. The key tasks of the manager seem to be most
generalizable in the form suggested by Kotter (1982, 1988).
That is that the manager leads the organisation by
generating and expressing an idea of where the organisation
needs to be going (agenda building). He/she liaises with
networks of people and influences them to help in
implementing those ideas (networking). Finally, the manager
ensures that the agenda items are implemented through a
21
variety of control, influence and disturbance handling
tactics (execution).
Process Features
1. Managers, particularly at chief executive levels, work
long hours. Carlson (1951) found that the directors in his
study worked between 8.5 and 11.5 hours a day. The general
managers in Kotter's study worked an average of nearly 60
hours per week. The work of chief executives extends beyond
the office into home and social environments (Elliot 1959)
and dominates their thinking, even when they are not
physically working (Mintzberg 1973, Carroll and Gillen
1987) .
Managers at lower levels work shorter but still sUbstantial
hours (Burns 1957, Horne and Lupton 1965). The middle
managers in Horne and Lupton's (1965) study, for example,
worked around 44 hours per week. The hours of work vary
between different organisations, work types and countries
(stewart 1988).
2. Managers deal with large amounts of work comprised of
diverse work demands. The work process is brief, intense,
fragmented and highly demanding (Carlson 1951, Guest 1956,
Dubin and Spray 1964, Mintzberg 1973, Kotter 1982, Cox and
Cooper 1988, Stewart 1988). Most managerial time is taken
22
up with day-to-day crises, interruptions and ad-hoc problem
solving (Hales 1986, Martinko and Gardner 1990). Guest's
(1956) study of foremen, for example, found that they were
involved in between 237 and 1073 separate incidents daily.
The pressure of work is such that even very senior managers
are unable to spend much time on formal planning (Hales
1986, stewart 1989). As a consequence managers emerge as
"intuitive responders rather than strategic planners"
(stewart 1982, p.90).
3. Manag,;rial work is very much a social process (Hosking
and Fineman 1990). Verbal interaction occupies between "two
thirds and four fifths" (Hales 1986, p.98) of the manager's
time (Burns 1954, 1957, Guest 1956, Horne and Lupton 1965,
Mintzberg 1973, 1989, 1990, stewart 1976, 1988, Fry,
Srivstva and Jonas 1987, Jonas 1987). The exact proportion,
pattern and difficulty of verbal contacts varies between
jobs (Dubin and Spray 1964, Kelly 1964, Mintzberg 1973,
Stewart 1976, Hales 1986). A.I though the manager's
interactions may range across hundreds or even thousands of
contacts (Kotter 1982) the majority of interactions are
lateral (Burns 1957, Dubin and Spray 1964, Horne and Lupton
1965, Mintzberg 1973, Stewart 1976, Hales 1986).
Much of this interaction involves the manager in attempts to
influence other people to do things, through brief face to
23
face conversations (Hales 1986). The manager has less
direct pow43r, in these relationships, than is commonly
supposed (Sayles 1979, Kotter 1982).
written communication, even in the form of hard information
such as reports and computer print-outs, receives less
attention 'than verbal communication (Mintzberg 1973, 1975,
1989, Daft, Sormunen and Parks 1988). By contrast managers
pay close attention to soft verbal information, such as
gossip and hearsay (Neustadt 1960, Mintzberg 1973, 1989).
The neglect of written material however, may not be as
complete as is implied by this research. There is evidence
that managers may use their t:ime out of working hours to
address more formal written materials '(Brewer and Tomlinson
1964).
4. Although largely ignored by managerial texts, feelings
and emotions play an important role in managerial work.
Managers exhibit the same range, "richness and poverty of
emotions" (Hosking and Finema.n 1990, p.595) as other human
beings. These feelings and emotions have a powerful impact
on their work behaviour and personal experience (Herzog
1980, Terkel 1985, Hosking and Fineman 1990). An ability to
access and respond to personal feeling states and emotions
is an important aspect of thc3 intui ti ve responses required
in managerial work (stewart 1982, Bennis 1989, Mintzberg
1989) .
24
5. The verbal, ad-hoc nature of managerial work can be
highly efficient, both in terms of a fragmented and highly
pressured internal environment and a strategic environment
which is discontinuous (i.e. variously interrupted, delayed
and speeded-up) and dynamic in nature (Mintzberg 1989). The
near constant interaction with people provides the manager
with the opportunity to form, test and modify agendas whilst
simultaneously developing the networks needed to implement
them (Brewer and Tomlinson 1964, Kotter 1982, Hales 1986,
Mintzberg 1990). The following quote from Kotter (1982,
p.166) is illustrative; allow the general managers
to react in an opportunistic (and highly efficient) way to
the broad flow of events around them, yet knowing that they
are doing so within some broader and more rational
framework. The networks allow terse (and very efficient)
conversations to happen; without them, such short yet
meaningful conversations would be impossible."
In the sense of developing agendas managers are guided by a
plan. However, it is not the formal plan outlined by the
classical writers but a looser mentally held grouping of
"flexible but often specific, intentions [formed] in the
context of daily actions" (Mintzberg 1990, p.165).
6. Efficient use of managerial agendas demands that the
manager's task performance be simultaneous, interactive and
holistic in nature (Mintzberg 1973, 1976,1989, Kotter 1982,
25
26
Weick 1983, Hales 1986). Managerial tasks are not performed
in a sequential, linear fashion but form an interactive
gestalt (Mintzberg 1990). Hales (1986, ,p.102) notes that
"managerial work is not the sequential execution of separate
activities but is often an artful, simultaneous synthesis of
inter-dependent activities ... There is both rapid commuting
between activities ... and the simultaneous execution of
discrete and separable activities, with the one activity
providing the context, even the opportunity, for carrying
out others".
An interactive view of managerial work is reflected in an
emerging European research focus (see for example Bouwen and
steyaert 1990, Brown 1990, Hosking and Fineman 1990). In
this work the management process is characterised as having
"a feel, a processual weave, a sense of actions, passions
and politicality which [give] it an interconnectedness and
texture (Fineman and Hosking 1990, p.573). Reference to the
connectedness, complexity, texture and context of the
organising process are typical of this work (see Hosking and
Fineman 1990).
7. Managerial work presents competing 'demands and
pressures. Much of the manager's work involves coping with
and reconciling the conflict, ambiguity, and cross-pressures
inherent in the job (Hales 1986, stewart 1989). The ongoing
r
interaction of conflict and compromise produces a work
process which is inherently political in nature (Brown 1990,
Hosking and Fineman 1990, see also content feature five
above) .
What are the Boundaries of Managerial Work?
The findings outlined above imply that managers can be
defined in terms of content and process characteristics that
form a distinct and exclusive part of the managerial
function. Unfortunately there has been no research which
has attempted to compare the work of managers with those of
non-managers (Hales 1986, stewart 1989). "In short, the
studies have not demonstrated that there is a bounded and
separable set of activities which may be called -
\managerial work' - and not merely activities which managers
have been shown to dOll (Hales 1986, p.109).
There is a sUbstantial sociological literature which asserts
that the characteristics associated with managerial work are
widely disseminated through other non-managerial occupations
(see Braverman 1974, Marglin 1976, Nichols and Beynon 1977
and storey 1980). The implication is that the managerial
function, rather than making a distinct and identifiable
contribution, acts as an ideologically linked status
justifying the inequitable distribution of organisational
benefits (Anthony 1977).
27
The absence of empirical evidence makes it difficult to
respond directly to such arguments, or to clearly specify
the parameters of exclusively managerial work. Nevertheless
it is possible to define the term manager in ways that
differentiate it from other social functions. Neither does
it seem unreasonable to assert that the features of
managerial work, while perhaps not exclusively managerial,
have sufficient specificity to bring further definition to
the managerial function.
WHAT CONSTITUTES EFFECTIVE VERSUS INEFFECTIVE MANAGEMENT?
In the previous section we looked generally at the
managerial function and attempted to define it semantically
and in terms of identifiable characteristics. In this
section we look at research and current research issues
relating to the characteristics and behaviours of
specifically effective versus ineffective managers
Previous Research
There has been little research to date on the differences
between effective and ineffective managers. Attempts to
link observed behaviour with effectiveness were strengths of
some of the critical incident studies of the 1950's (see
Flanagan 1951, 1952, Kay 1959). Effectiveness was also a
focus in the early studies of foremen (Guest 1956, Jasinski
28
1956, Roach 1956, Kay 1959). It is not until comparatively
recently however, that the issue of effectiveness has again
received serious attention (Morse and Wagner 1978, Luthans,
Rosenkrantz and Hennessey 1985, Martinko and Gardner 1990).
The later studies provide valuable insights into the
behaviours associated with positive performance outcomes.
They indicate that effective behaviour varies between
organisations and management levels (Morse and Wagner 1978,
Luthans et al 1985). They also highlight the impact of the
environment on managerial behaviour (Martinko and Gardner.
1990). These studies however, are but a small beginning in
an area which has been substantially neglected.
For the most part o u ~ knowledge of managerial work is not
grounded in any concept of effectiveness. While we have a
very sUbstantial literature on managerial practise we have
very little literature which attempts to distinguish between
effective and ineffective managerial practise. stewart
(1982) for example, does an outstanding job of highlighting
the diversity of managerial work and the choices available
to the manager in defining that work. She provides no
information however, as to the efficacy of the choices made,
in terms of outcomes, or of the belief systems that' actuate
those choices.
29
Most of the management literature, including studies that
claim to describe effective management (see for example
Sayles 1979) do not even define the term effectiveness, far
less attempt to operationalise the concept. This
definitional inadequacy reflects the lack of integrating
theory, in both the management literature (Martinko and
Gardner 1985, Hales 1986, Stewart 1989) and in the related
organisational effectiveness literature (Goodman, Atkin and
Schoorman 1983, Lewin and Minton 1986). It also reflects a
failure in the management literature to develop consistent
terms and cat.egories (Hales 1986).
The lack of attention to effectiveness is a serious flaw in
the literature which has been highlighted in all of t h ~
recent reviews (Martinko and Gardner 1985, Hales' 1986,
Stewart 1989). As indicated by Luthans et al 1985, p.257)
there is a need "to go beyond asking what managers really
do" and ask instead "what do successful managers really do?"
This is one of the most important requirements of future
management research.
Defining Managerial Effectiveness
Figure 2:1 outlines a person-process-product model of
managerial effectiveness drawn from Campbell, Dunnette,
Lawler and Weick (1970). The model indicates that
managerial effectiveness can be defined in terms of
30
FIGURE 2:1
A MODEL OF MANAGERIAL EFFECTIVENESS
Individual Characteristics Individual outcomes
.......
v-
.....
II""
(Aptitudes, Values, Preferences etc) Behaviour (Profit, Survival
(Intelligence)
6
(Productivity etc)
/
Internal/External Organisation Environment
Person Process Product
31
individual characteristics, individual behaviour and
organisational outcomes. The term characteristics refers to
the personal qualities and traits that are required for
managerial success. Such characteristics have been
exhaustively documented in the trait research (see for
example Ghiselli 1971, Stogdill 1974, Bowen and Attaran
1987). They include intelligence, aptitudes, knowledge,
temperament and personality. The term behaviour refers to
the way managers conduct themselves, in their observed
actions towards others and in response to various work
situations.
This is an interactive model in that it assumes that the
person, process, product dimensions will influence each
other, with the primary concern being the impact of
.
managerial characteristics and behaviours on organisational
outcomes (e.g. level of profit, productivity, efficiency).
As indicated by this model and recent research, the pattern
of individual characteristics and behaviours that lead to
desired outcomes is contingent on the internal
organisational environment (e.g. its tasks, functions,
policies, procedures, conditions, resources) and the
external environment (e.g. uncertainty, market
characteristics). Individual characteristics and patterns
of behaviour that are effective in one context may not be so
in another (Morse and Wagner 1978, Luthans et al 1985). The
of the manager is determined by the "degree of
fit" (Hales 1986, p.111) between the characteristics and
behaviours of the manager and the demands of the particular
job situation.
The model implies that a definition of managerial
effectiveness should fulfil at least two requirements.
First, it must link the characteristics and behaviours of
the individual with desired organisational outcomes. Second
it must acknowledge that the pattern of effective behaviour
wi-II vary across different jobs, bosses, organisations and
environments and in response to the characteristics of the
individual manager (Campbell et al 1970, Fiedler 1974, Morse
and Wagner 1978, Hales 1986).
32
Hales (1986, p.88) notes a recognition of contingency in a
number of effectiveness definitions (see for example
Campbell et al 1970, Morse and Wagner 1978, Boyatzis 1982)
in that they all denote "the extent to which what managers
actually do matches what they are supposed to do". Hales
later notes (p.111) that what managers are supposed to do
will depend on the expectations, "tasks and functions"
surrounding a particular management job.
with the constraints outlined above in mind and drawing from
existing definitions we may define the effective manager as;
One who optimises the long term functioning of the
organisation by engaging in the behaviours best fitted to
the particular internal and external environment in which
they manage and to their own characteristics and
preferences.
The term optimises is used rather than maximises in
deference to the Seashore and Yuchtman (1967) argument that
maximisation of outcomes such as profit or growth would
generate imbalances which could be dysfunctional. The term
functioning derives from Campbell et al's (1970) definition.
It acknowledges a concern, both with performance ,outcomes,
for example, survival, profit and productivity and with
outcomes related to the internal characteristics of the
organisation, for example level of participation,
readiness and morale (Mahoney 1967, Campbell
1977, Lewin and Minton 1986).
33
Measuring Managerial Effectiveness
The definition outlined above, implies that descriptions of
the characteristics and behaviours of effective managers
need to work at two levels. First, they must be linked to
some measure of the outcomes that optimise long term
organisation functioning. they must identify the
managerial characteristics and behaviours that are most
efficacious in obtaining those outcomei in the specific
environmental context of the manager, We will look first at
the issues surrounding outcome measurement. We then turn to
the development of categories against which the outcome
measures can be compared.
Objective versus Subjective Outcome Measures
The selection of criteria for use in outcome measurement has
a long and chequered history (Smith 1976, Nathan and
Alexander 1988). Outcome criteria range between hard
objective (for example production quantity) and soft
sUbjective (for example supervisory rankings) .
A variety of objective criteria have been employed. Morse
and Wagners' (1978, p.31) study, for example, used "economic
end result data such as return on investment and budgeted
versus actual costs". Martinko and Gardiner's (1990) study
of school principals used measures of student performance on
minimal competency and standardised achievement tests.
Other objective criteria include profit, sales, rates of
return, production quantity, production quality,
absenteeism, productivity, accidents, staff morale and
turnover (Campbell et al 1970, campbell 1977, Boyatzis 1982,
Lewin and Minton 1986).
Objective criteria are frequently deficient in that they
ignore important aspects of the job. Production output, for
example, is only one aspect of a first level supervisor's
34
job. Such objective performance criteria do not acknowledge
the impact of the manager's behaviour on internal
organisational or. unit characteristics, such as morale and
satisfaction (Mahoney 1967, Campbell et al 1970, Lewin and
Minton 1986).
Perhaps more significant, is the potential for objective
measures to be contaminated by factors beyond the manager's
control. As implied in Figure 2:1, the "human, financial,
and material resources" (Campbell et al 1970, p.10S)
available to the manager and the conditions of the external
environment can impact dramatically on their outcomes.
objective criteria do not account for the impact of such
uncontrollable factors on the perceived effectiveness of the
manager (Campbell et al 1970, Nathan and Alexander 1988).
The most widely used subjective outcome criteria are global
rankings and ratings, often conducted by the manager's
superior (see for example Mahoney, Jerdee and Nash 1960,
Morse and Wagner 1978, and Martinko and Gardiner 1990).
Such criteria are useful in that they provide an overall
indicator of a manager's effectiveness, in relation to their
contribution to organisational functioning. They can also
be expected to cover a range of managerial behaviours
performed over time and are less likely to suffer from
deficiency problems, than objective criteria. Observers
inside the organisation are likely to be aware of what the
35
individual manager has contributed, herice sUbjective
criteria are less likely to be contaminated by external
factors beyond the manager's control.
The major weakness of subjective criteria is the inability
of the researcher to discern the extent to which the
judgement of the observer is contaminated by observational
errors such as halo, central tendency, leniency, limited
observations and bias relating to factors such as age, sex,
education, appearance and race (Campbell et al 1970, Nathan
and Alexander 1988). A ~ e l a t e d weakness of superior only
rankings/ratings is that other members of the ratee
manager's constituency (for example, peers, subordinates,
clients) are not included. The perceptions of superiors may
not be shared by other equally perceptive members of the
manager's constituency.
A variation on ranking/rating methods are salary or
promotion indices corrected for age or length of job tenure
(Hall 1976, McCall and Segrist 1980, Luthans et al 1985). A
further variation on this approach is to define the manager
as effective or successful by virtue of the fact that they
have made it into a top management position (Luthans et al
1985, Cox and Cooper 1988). Such criteria retain the
advantages of global rankings and have the additional
benefit of representing the perspectives of a number of
superiors accumulated over the manager's time in the
36
organisation. These perspectives presumably reflect the
manager's capacity to consistently contribute to the
functioning of the organisation.
These criteria are reliant on the capacity of the promotion
system to accurately reflect the effectiveness of the
manager. It will always be difficult for the researcher to
know to what extent the process is contaminated, for example
by level of competition, extent of observation of the
managers behaviour and the kind of judges the manager has
had over his/her tenure in the organisation. Hence, while
the pooling of perspectives, implicit in the promotion
process, reduces the likelihood of observational error it
increases the potential of contamination from factors beyond
the control of the individual manager. The promotion
process is also weighted toward the perspective of superiors
rather than subordinates and other members of the manager's
constituency. In New Zealand some organisations
(particularly in the public sector) are informally known for
their capacity to systematically promote less effective
people. This s6rt of informal understanding casts doubt on
the use of promotion indices as output criteria in such
organisations.
There are no definitive answers as to the relative efficacy
of objective versus sUbjective output criteria. Nathan and
Alexander's (1988, p.531) research, for example, found no
37
support for the "assumption that 'objective' measures of
performance are more predictable than sUbjective
evaluation" .
Hales (1986 p.108) contends that subjective criteria may
actually be more appropriate in managerial settings than
"some absolute, objective, benchmark". Subjective criteria
are seen by Hales, as better adapted to the varied and
contingent nature of managerial work and effectiveness.
The uncertainty of current organi.sational and managerial
environments does lend support to the use of sUbjective
effectiveness criteria. The ability of subjective criteria
to focus over time, on specific managers, in specific
contexts and to limit contamination from external factors,
may give such criteria an advantage in studies of managerial
effectiveness. Furthermore, there is the concern that
objective measures, however well constructed, may be
meaningless if based on unreliable data.
There are indications that the distinction between objective
and subjective criteria is somewhat artificial. Both are
ultimately reliant on human judgement (in the case of
objective criteria in the choice of performance standard).
In this sense both are subjective in nature and it is
perhaps not surprising to find similarity in
predictabilities (Jaques 1976, smith 1976, Nathan and
Alexander 1988). Nathan and Alexander (1988, p.530) for
38
example, .found the overlap between "observed validity
distributions from ... subjective ratings and objective
production quantity ... of ~ u c h a magnitude that little
meaningful differentiation between the use of these criteria
could be determined".
Overall we may conclude with Nathan and Alexander (1988,
p.533) that the selection of objective versus subjective
output criteria is not "as serious a problem as has been
generally assumed". Ideally multiple criteria would be
employed (Goodman and Pennings 1977, Morse and Wagner 1978,
Heneman 1986, Martinko and Gardner 1990). Where objective
criteria are not available sUbjective criteria can be
expected to provide similar predictability and may, in fact,
be better adapted to' studies of managerial effectiveness.
Developing categories of Effective Managerial Behaviour and
Characteristics
Having established criteria against which managerial
characteristics and behaviours can be evaluated we now need
a method for identifying and categorising such
characteristics and behaviours. As we discussed above, one
of the key constraints in developing effectiveness criteria
lies in the varied and contingent nature of managerial work
and effectiveness. This same variety and contingency is
39
also a constraint in developing characteristic and
behavioural categories.
Given the diversity of managerial work it is desirable to
avoid the use of preconceived categories, frameworks and
perceptions. There is always the danger of imposing a
generality which is not relevant and/or missing some aspect
which is particular to a specific research situation. The
need is to rely on "the managers themselves rather than on
psychologists to choose the appropriate definitions,
wordings and format" that categorize their work (Campbell et
al 1970, p.479). Not surprisingly, (Hales 1986, p.93)
reports a recent "shift away from the measurement of
managerial jobs across pre-formed categories toward the
discovery of categories".
This shift is not in evidence in recent studies of
managerial effectiveness. Most of the recent studies of
managerial effectiveness make use of pre-formed frameworks
(see Morse and Wagner 1978, Luthans et al 1985, Martinko and
Gardiner 1990). This is fine in studies that are attempting
to compare existing behavioural frameworks (such as
Mintzberg's roles) with effectiveness measures taken across
diverse organisational settings. It is less useful in
studies which are attempting to discover effectiveness
characteristics and behaviours, particularly those that
relate to specific organisational settings.
40
The research question guiding this study and the paucity of
existing answers to that question implies a process of
discovery which is more exploratory and inductive in nature
than it is deductive. It is the earlier, rather than later
studies which evidence such an approach. Flanagan (1954)
developed the critical Incidents Method in which qualified
observers are asked to report examples of particularly
effective or ineffective behaviour. Once the incidents are
collected they can be mined for behavioural categories for
use on rating forms. This is a useful approach which
appeared frequently in the literature of the 1950's
(Flanagan 1 9 5 ~ , 1952, Kay 1959).
A slightly different method was adopted by Roach (1956). In
this study managers were asked to write "a brief essay
describing the behaviour of the best and poorest supervisor
they knew" (Roach 1956 p.488). These essays, seventy in
all, were then content analysed to produce a "checklist-type
questionnaire ... in which supervisors could be described by a
five point scale depending on the applicability of the
statement to the supervisor being described" (Roach 1956
p.488) .
Campbell et al (1970) describe a method for developing job
behaviour observation scales which involves five workshop
discussion sessions with experienced managers from target
organisations. These workshops are used to develop
41
behavioural dimensions and critical incidents which are
sorted into job behaviour scales. stewart and stewart
(1981a) developed the items for their performance
questionnaires by holding brainstorming sessions with
lipeople from the personnel department, outside experts,
behavioural scientists, interested line managers and so on"
(p.84,85). In all of these approaches, the researchers have
attempted to allow the managers to speak for themselves,
rather than imposing their own frameworks. These methods
are much more likely to capture the flavour of distinctive
research situations than are applications of pre-formed
categories. However, they are not without limitations.
In all of these methods the researcher collects a series of
written incidents, essay examples or group perspectives. In
some of these methods the process of collection proceeds
through the me.dium of a pre-formed interview question
format. The greater the reliance on such a pre-formed
format the greater the likelihood of observer bias entering
the research. Use of such formats also increases the
likelihood of respondents offering espoused theories as
opposed to identifying dimensions that are of real
significance to themselves (Ginsberg 1989). In this case
the respondent is cued by the perceived requirements of the
question framework or of the researcher.
42
Having gathered the data the researcher must ,give definition
to what is discovered and hence to the categories and models
that emerge. In the critical incident and essay studies of
Flanagan 1951, Roach 1956 and Kay 1959) the recorded
observations must be "evaluated, classified and recorded"
and finally "summarized and integrated" (Flanagan 1952,
p.384,385). with the approach suggested by Campbell et al
(1970), the effectiveness dimensions and categories that
emerge from the workshop process require definition and
structuring on. the part of the consultant or researcher.
This is also the case in the brain storming process outlined
by stewart and stewart (1981a) and invoked concern on the
part of the author's regarding the impact of observer bias.
The collection of raw data in the form of aggregated
incidents, essays or group perspectives can present the
researcher with a difficult task in the
definition/classification phase. There is frequently a need
for considerable interpretative input on the part of the
researcher, increasing the potential impact of observer
bias. Some of the insights expressed by respondents may
also be lost. In past research a high proportion of the
ideas identified by these methods have failed to
subsequently discriminate between managers identified as
effective and ineffective (stewart and Stewart 1981a,
1981b). As a consequence the methods outlined above, are
not particularly productive in terms of the usable
43
effectiveness items produced (stewart and stewart 1981a,
1981b) .
Concerns of this nature led to the adoption, for industrial
use, of a clinical procedure called the Repertory Grid
Technique. The Repertory Grid Technique is a semi-
structured interview process which can be used to explore
the ideas and frameworks used by individuals in categorising
managerial effectiveness. It has a number of advantages
over other qualitative methods.
Most important of these, for our purposes, is that it is a
technique largely free of observer bias. It elicits the
m ~ a n i n g s held by the respondents themselves rather than
imposing the frameworks and- cognitive construction systems
of the researcher (stewart and stewart 1981a, 1981b, Crow
1988, Ginsberg 1989). The data elicited in the RepertQry
Grid interviews falls out in a series of bi-polar
descriptors. These data have the advantage of being clearer
than most qualitative data bases and are therefore easier to
categorise and prepare for further analysis.
The Repertory Grid is a highly efficient technique,
generating much higher amounts of usable data than
comparable qualitative techniques (stewart and stewart
1981a, Dunn and Ginsberg 1986, Ginsberg 1989). stewart and
stewart (1981a) report a productivity increase (in terms of
44
usable categories) of around 500% when they replaced their
brainstorming techniques with Repertory Grid approaches.
Finally, it is a readily replicable technique producing data
which can be analysed and validated using computer driven
statistical analysis (Bell 1987).
These advantages are confirmed by applications of the
Repertory Grid Technique in a variety of organisational
settings. It has been used for example, in research on
management information systems (Stabell 1978), occupational
stress (Crump, Cooper and smith 1981) managerial performance
(Stewart and Stewart 1981a, 1981b), organisation structure
(Wacker 1981), organisation innovation (Dunn and Ginsberg
1986) and competitor and portfolio analysis '(Walton 1986,
Ginsberg 1989). These studies indicate that the Repertory
G ~ i d Technique is an ideal method for developing
characteristic and behavioural categories of the sort
required in this study. The Repertory Grid Technique
provides one of the best exploratory research methods
currently available and hence has been adopted for this
study. The method and its application to this study are
discussed fully in chapter three.
45
HOW DO THE CHARACTERISTICS AND BEHAVIOURS OF EFFECTIVE
MANAGERS VARY ACROSS DIFFERENT MANAGERIAL LEVELS?
The literature indicates that organisational level does
influence the characteristic and skill requirements of
managerial work (see fqr example Hemphill 1959, Dubin and
Spray 1964, Mahoney, Jerdee and Carroll 1965, Child and
Ellis 1965, Horne and Lupton 1965, Thornton and Byham 1982,
Pavett and Lau 1983, Luthans et al 1985). By the term level
we mean a grouping of staff of roughly equal status and
. responsibility. There is reasonable agreement within this
literature concerning patterns of variation between
management levels.
Katz (1974) argued that effective management rests on three
central skills. These are conceptual skills, technical
skills and human skills. In conformity with K a t z ~ s (1974)
reasoning most studies report an increased emphasis on
longer range conceptual tasks and skills with movement up
the managerial hierarchy (Hemphill 1959, Mahoney et al 1965,
Haas, Porat, and Vaughan 1969, Tornow and Pinto 1976, Pavett
and Lau 1983, Dakin, Hamilton, Cammock and Gimpl 1984,
Luthans et al 1985, McLennan, Inkson, Dakin, Dewe, and Elkin
1987) .
Conceptual skills, as defined in these studies, involve two
primary dimensions. The first relates "to the sensing of
46
the organization as a whole and the total situation relevant
to it" (Barnard 1938, p.235). The second, related
dimension, involves "systematic long range thinking and
planning" (Hemphill 1959, p.59). The increase in emphasis
shown in these studies is supportive of the proposition that
"events become more spontaneous and unplanned as jobs move
down the managerial hierarchy" (Martinko and Gardiner 1990,
p.347 see also Mintzberg 1973).
Jaques (1976) offers a more sophisticated typology' of the
distinctive types of conceptual thinking required at
different organisational levels. Jaques (1976, p.153) uses
the concept of the "level of abstraction" required in the
thinking of the manager. These range from concrete thinking
"carried out in direct physical contact with the output" for
by a first level supervisor (Jaques 1976, p.144),
through to highly abstract processes based on "unconscious
intuition, with a complex of apparently unrelated facts and
figures" (Jaques 1976, p.151). These much higher levels of
abstraction are characteristic of high level managerial jobs
and decisions with a very long time horizon.
Most, but not all (see Ramos 1980, Pavett and Lau 1983),
commentators report a decreasing emphasis on
specialist/technical knowledge and skills with movement up
the managerial hierarchy (Fayol 1949, Barnard 1938, Hemphill
1959, Thornton and Byham 1982, Dakin et al 1984, McLennan
47
al 1987). Corresponding with this shift is the need to
leave behind more specialist, technical roles and adopt a
more generalist approach at senior managerial levels
(Mahoney et al 1965, Dakin and Hamilton 1986).
Some distinction needs to be made between the various
definitions of technical knowledge and skills. The Katz
(1974, p.91) definition is based around "an understanding
of, and proficiency in, a specific kind of activity
particularly one involving methods, processes, procedures or
techniques". Involvement in specialist activity of this
kind would seem likely to decrease with movement up the
managerial
Kotter (1982, p.134) provides a broader definition of
knowledge as based on a "detailed knowledge of the
business and organization and good solid relationships with
the large number of people upon whom the job makes him
dependant". Kotter's definition involves broad industry
experience, knowledge and networks. The need for a
technical grounding of this nature limits the organisational
transferability of top managers (Kotter 1982, Dakin et al
1984, Whitely 1989). Technical knowledge of this sort is
undoubtedly important, even at very senior management levels
(Kotter 1982, 1988). We may add a third definition of
technical knowledge, based on the need for the managerial
technical skills that are offered in courses such as the
48
Masters of Business Administration (MBA). These include for
example, financial analysis, marketing, computing and human
resource m a n a g e m e n ~ . One of the key purposes of the MBA
process is to provide potential general managers with a
generalist grasp of these skill areas. We may assume that
such a generalist understanding becomes more important as
the manager advances up the managerial hierarchy.
The human skills dimension is important at all managerial
levels (Katz 1974, Pavett and Lau 1984, Dakin et al 1984,
Bonama and Lawler 1989). The need for human skills however,
appears to become "proportionally, although probably not
absolutely, less" (relative to other skills) as the manager
advances up the hierarchy (Katz 1974, p.95, see also Pavett
and Lau 1983).
49
The nature of the required people interactions also changes
between managerial levels. At lower levels managers are
primarily involved in internally orientated supervisory
tasks such as directing, leading and developing subordinates
(Hemphill 1959, Mahoney et al 1965). At more senior levels
the people contacts become more complex and externally
oriented (Hemphill 1959, Mintzberg 1973, Alexander 1979,
Paolillo 1981, Pavett and Lau 1983, Luthans et al 1985).
A related area concerns managerial skill and activity in
organisational politics (Dalton 1959, Fletcher 1973). There
has been little research on variations in the between level
emphasis on this area. Pavett and Lau (1983) found no
significant difference in the perceived requirement for
political skills between management levels. Luthans et al
(1985) by contrast, found that first line and middie
managers engaged in significantly more political
than did the top managers in their sample.
The existence and patterns of variation in the between level
characteristics and skill requirements of managerial work
are well illustrated by the literature. The primary
deficiency of this literature, as with that related to more
general descriptions of managerial work, is its failure to
describe level by level differences between specifically
effective and or ineffective managers. The literature
offers no explicit guidance about the ways in which the
skills and characteristics needed to be effective in
management vary across organisational levels. This is
obviously a useful area for research and is therefore one
aspect of this study.
CONCLUSION
In this review we have acknowledged the influence (and the
constraints on that influence) of managers on key
organisational and societal outcomes. We have also, along
with Martinko and Gardner (1985), Hales (1986) and stewart
50
(1989), noted the paucity of research examining
effectiveness in management. Few studies have attempted to
discover the characteristics and behaviours that distinguish
effective from ineffective managers. Furthermore, there is
an almost complete absence of published research exploring
the nature of managerial effectiveness across organisational
levels.
with these issues in mind we have proposed the following
research question; "What are the characteristics and
behaviours of effective versus ineffective managers and how
do these characteristics and behaviours vary between
different managerial levels." The methods used in
addressing this question along with the results and their
implic.ations are discussed in subsequent chapters.
Chapter three describes the Grid Technique in more
detail. Chapter four discusses data gathering and sampling
approaches. Chapters five and six define the major scale
and factor categories relating to the characteristics and
behaviours of most and least effective managers in this
study. Chapter seven tests hypotheses relating to variation
in the characteristics and behaviours of most and least
effective managers at different managerial levels. Chapter
eight summarises the major findings and discusses their
implications for managerial practise, teaching and research.
51
CHAPTER THREE
REPERTORY GRID TECHNIQUE
INTRODUCTION
In chapter one we outlined our research question as; "What
are the characteristics and behaviours of effective versus
ineffective managers and how do these characteristics and
behaviours vary between different managerial levels."
Addressing this question presents two central
First it requires a means of identifying behaviours and
characteristics which are representative of effective and
iheffective managers. Second it requires effectiveness
criteria against which the the validity and relative
importance of the behaviours and characteristics can be
established. In this chapter we focus on the identification
of behaviours and characteristics and the role of the
Repertory Grid Technique in this process. The effectiveness
criteria used in this study are discussed in chapter four.
52
In chapter two we noted the varied and contingent nature of
managerial work. We found that the process and content of
managerial work can vary significantly across different
levels, job types, organisations, environments, cultures and
even between different managers performing similar jobs.
This variat.ion demands researc:h methods that allow
individuals;, not researchers, to define the characteristics
and behavieturs which relate to 'managerial effectiveness in
their particular situation.
This has nett been the case in recent research on effective
and ineffective managers (see for example Luthans et al
1985, Martinko and Gardiner 1990). As we saw in chapter
two, most clf these studies have used frameworks and
instruments developed in othe::::- research settings. The
danger of such research approaches is that they bring a
.definition and structure to the research setting which may
be inappropriate. ~ t is difficult to know in such an
approach hClw much the finding::; reflect the perspectives of
the respondents and how much they reflect those contained in
the method or the researcher. This same difficulty was also
present in the earlier more qualitative studies of
managerial effectiveness (see for example Flanagan 1951,
Roach 1956, Kay 1959, stewart 1981a, 1981b), with their
heavy reliance on researcher -evaluation and classification.
The desire in this study was to allow individuals to speak
for themselves, within the context of their particular
circumstances, without the distorting influence of
instrument or researcher pre-conceptions. The Repertory
Grid Technique addresses this need well. Its use allows
53
respondents to describe their own managerial worlds, with
their attendant categories, with minimal interference from
the researcher. The Repertory Grid is to
this study. In this chapter we describe the technique and
the principles that underly it.
BACKGROUND TO THE REPERTORY GRID
The Repertory Grid Technique (R.G.T.) was developed by
George Kelly for use in clinical practise (Kelly 1955).
Kelly's development of the R.G.T was motivated by two
primary concerns. The first related to the impact of
observer bias on the diagnosis of clinical patients. Kelly
sought an approach which would allow the perspectives of the
patient to 'emerge without the distorting influence of the
clinician's training and perspectives. At the therapeutic
level the need was for an approach which would allow the
therapist to explore with the client rather than impose
externally derived perspectives.
The second concern related to the then current obsession, in
the field of psychology, with statistically based studies of
large masses of people. Kelly felt that individual
differences were being neglected in this research and sought
a clinical approach that would allow individual clients to
speak for themselves, rather than being categorised on the
basis of large sample norms.
54
Kelly's emphasis on idiographic research approaches draws
from quite radical assumptions regarding human nature.
These assumptions parallel emerging views of managerial work
(see chapter two) as simultaneous, interactive and holistic
in nature. The theory (known as Personal construct Theory)
underlying the R.G.T. refutes the notion of a static and
therefore predictable and controllable human nature. People
are seen as "a form of motion" (Bannister and Fransella
1986, p.63) continually experimenting, evolving, and
reconstructing within their life experience.
The attempt, implicit in orthodox psychology, to "fathom the
nature of humanity" is meaningless when placed in the
context of construct theory. It is the individual, not the
scientist who explores, defines and continually redefines
the issues of life and humanity. In this sense individuals
are scientists and experimenters, continually seeking to
"understand their own nature and the nature of the world and
to test that understanding in terms of how it guides them
and enables them to see into the immediate and long term
future" (Bannister and Fransella 1986, p.8). People will
vary from each other in the way they construe life and life
events. As a consequence "each of us lives in what is
ultimately a unique world, because it is uniquely
interpreted and thereby uniquely experienced" (Bannister and
Fransella 1986, p.10). Construct theory argues for an
55
approach that accompanies the individual in their unique
construction of meaning, rather than imposing externally
derived models and frameworks that reflect the perspectives
of some other person or body of understanding.
Kelly's concerns echoed our own need to establish
effectiveness categories that reflect the perspectives of
specific individuals and research settings. Similar
concerns, albeit more narrowly focused, have been expressed
by others who have adopted the R.G.T. stewart and stewart
(1981a, 1981b) use the R.G.T to tailor-make management
development around the needs of individual organisations,
rather than assuming that the training needs of managers are
the same in all organisations. Crump, Cooper and smith
(1980,1981) adopted the R.G.T in preference to widely used
pre-designed health and behavioural questionnaires. The
R.G.T was seen as more effective than pre-designed
instruments, in involving respondents and developing a
representative data base.
Because the Repertory Grid interview process works with
elements rather than a schedule of interview questions, its
reliance upon researcher input and interpretation is
minimal. It is perhaps not surprising that researchers
active in the use of the R.G.T claim that the problems of
observer bias and Hawthorne effects, inherent in so many
other research approaches are almost completely absent with
56
the R.G.T (see Bannister and Mair 1968, Crow 1988, Ginsberg
1989) .
PERSONAL CONSTRUCT THEORY
The R.G.T is based on Kelly's Personal Construct Theory (see
Kelly 1955). Implicit in this theory is the idea that
people need to make sense of their environment and life
experience. The world of the infant, to paraphase William
James, is one of "buzzing, blooming, confusion". This
confusion is reduced by the maturing person through the
development of what Kelly calls personal construct systems.
The individual's personal systems the
ways in which they construe the people, objects and ev
7
nts
they encounter in their life experience. The individual's
psychological processes are in turn "channelled by the ways
in which he or she successively construes events" (Bannister
and Fransella 1986, p.63). It is the personal construct
system that guides the individual in their search for
meaning and in their attempts to anticipate and understand
future life events.
57
The individual's construct system is made up of a series of
dicotomous constructs. Kelly (1955, p.61) defines a
construct as "a way in which at least two elements are
similar and contrast with a third". Constructs are bi-polar
in nature, for example; light versus dark, happy versus sad,
strong versus weak. The individual's personal construct
system allows a series of bi-polar comparisons to be made,
through which they understand their environment by
simultaneously noting similarities and differences and by
searching for commonalities within diverse events (Easterby-
smith 1980). Individual constructs form part of a
hierarchical system in which constructs are linked in
subordinate and superordinate relationships. For example,
for some individuals, the construct sports car versus saloon
car might be subordinate to the construct good car versus
bad car. Both constructs might in turn fall under the
mobile side of the mobile versus immobile construct.
An individual's construct system may contain thousands of
such groupings, each with hundreds of constructs. The
construct employment (versus work), for example, might have
hundreds of types of paid activity grouped under it. This
allows the individual to handle a whole range of
constructions around the theme of employment. The linking
of these construct groupings provides the individual with a
complete system for understanding and dealing with the
people, objects, and events that confront them.
The accuracy of an individual's personal constructs are
constantly being assessed on the basis of their past
predictions relative to actual current outcomes (i.e. their
success in anticipating events). Personal constructs are
58
thus involved in an ongong validation and modification
process on the basis of current feedback. For example, a
manager who his/her efectiveness as based around
maintaining a distance from staff and emphasizing formal
authority will face a major challenge should the culture of
their organisation emphasize personal contact and team based
approaches. The failure of such a construction to produce
the anticipated success will challenge the validity of
his/her construct system and may lead to a change in the way
managerial effectiveness is construed. The concept that
emerges from this process is of an individual in a state of
near constant change, never quite the same from one moment
to the next. It is also in this process that we see the
individual as the scientist and the inquirer, constructing
hypotheses and modifying them in response to the results the
experiments have generated.
59
The people, objects and events that provide the focus of an
individual's personal constructs are called elements. The
constructs are ways of understanding the elements that
appear in the individual's life experience. constructs are
not passive labels but are an active means of evaluating and
discriminating between the elements that enter the
individual's life experience (Fransella and Bannister 1977,
Bannister and Fransella 1986).
60
Just as individuals are unique, so will personal construct
systems be unique in many ways. The researcher must seek to
understand the respondent's unique perspective on both poles
of their constructs. "Each construct is seen as a dichotomy
and the two opposing poles are individual and personal to
the construer. For example and could be
opposites for many people, but for some people might
be the opposite of (Crow 1988, p.l).
The constructs of different individuals will also have "many
similarities in content and structure due to a common
sensory and cognitive system, and a pool of common knowledge
that has been accumulating for thousands of years" (Lim
1984, p.28, see also Slater 1977). The R.G.T offers the
dual potentiality of pooling information held collectiveli,
while at the same time revealing distinctive individual
characteristics. In the case of this research, for example,
it provides a means of tapping and pooling the collective
views of a large number of respondents on management
effectiveness. At the same time it acknowledges and
explores the unique insights of specific individuals and
situations.
APPLICATION OF THE REPERTORY GRID TECHNIQUE
A variey of procedures are available for eliciting
respondent constructs. All of the procedures involve a
comparison of elements in terms of the respondent's personal
constructs. In this research we used the method of triads.
This method proceeds as follows;
i. Element Elicitation
The researcher or clinician will select a set of elements,
sometimes in conjunction with the respondent (or client).
An element is an event, person or object in the domain under
consideration. The elements are chosen to represent the
specific domain that the researcher and will
explore. For example, if the field of study was personal
relationships the elements might be; my wife, my best
friend, a person I dislike. It is in the choice of element
that the researcher has the greatest influence in the
interview process. Care needs to be taken to ensure that
the element set is representative of the domain under
consideration.
Elements are usually presented in groups of three (triads).
There is nothing sacred about the use of triads however and
dyadic presentations appear equally successful. The element
61
sets often include both positive and negative elements so
that both poles of the respondent's construct system are
explored. In the example given above, the elements my wife
and my best friend would act as positive elements with a
person I dislike acting as the negative. This allows both
poles of the respondent's construct system to be explored.
Having established an element set representative of the
domain of interest, the R.G.T commences with a preliminary
outline of the purpose of the interview and an assurance
that the responses will be held in complete confidence. The
respondent is then given a number of blank cards on which
they write the elements around which the interview will
revolve.
stewart and stewart (1981a, p.87) provide the following
example of a list of elements designed to explore the way
managers construe their own work activity.
1. An event where you feel you have performed well.
2. An event where you feel you failed to live up to your own
expectations.
3. An event which was important but which came as a bit of a
surprise.
4. A routine event that you enjoy.
5. A routine event that you dislike.
6. An important event requiring mainly managerial skills.
62
7. An important event requiring mainly technical/managerial
skills.
8. Another event where you feel that you performed w e l l ~
9. Any other event that is an important part of your work.
At the start of the interview respondents are asked to think
about each element in turn and to write on cards a specific
example of each of the elements. For example, in responding
to element one "an event where you performed well" the
respondent might write "the sales presentation I gave at
last months conference".
ii. Construct Elicitation
The cards are numbered to correspond with the element they
represent and in the triad method, the elements are
presented to the respondent one set at a time. The
interviewer commences by requesting the respondent to
consider the elements described on cards one, two, and three
and asks "I would like you to tell me one way in which any
two of these events are similar but different from the
third". In the case of the stewart and stewart (1981a,
p.87) example the respondent replied "Well, Planning and
Travelling are both solitary activities, but Selection
Interviewing involves other people". The interviewer then
recorded; Solitary ----- Done with others. Thus the first
construct emerged.
63
In the process of the interviews the interviewer will draw
comparisons between triads involving all or a number of the
elements. These comparisons will produce a series of bi-
polar constructs providing a full picture of the way in
which the respondent construes the subject being researched.
Appendix One provides an example of the constructs elicited
by an interview conducted as part of this research on
effective and ineffective managers. In eliciting constructs
the type of element is important. Our experience indicates
that the use of people or objects as elements proves easier
in construct elicitation than events. Respondents have more
difficulty generating constructs when abstract events are
used, rather than more con'crete elements such as people or
objects.
iii. Laddering Up and Down
As mentioned above, individual constructs exist in a
construct hierarchy. Construct relationships can be further
explored using Hinkle's (1965, 1970) laddering technique.
To ladder-up the construct hierarchy the interviewer asks
questions such as "Why is that important?" To ladder-down
the interview asks questions like "Why is that?", "What are
the implications of that?", "Tell me more about that?" As
an example, lets imagine the respondent was presented with
the following three elements.
64
1. The car I would most like to own.
2, The car I would least like to own.
3. Another car I would really like to own.
The respondent may record Lamborgini Diablo as element one,
with Lada 1300 and Porsche 911SC, as elements two and three
respectively. In response to the request "I would like you
to tell me one way in which two of these cars are similar,
but different from the third", the respondent may reply
"Well two of these are sports cars while the other one is a
saloon."
Laddering up to superordinate constructs in the hierarchy
involves the question "You mentioned that two of these cars
are sports cars and one is a saloon; why is that important?"
The respondent might then say "Well sports cars are exciting
saloon cars are boring". Exciting---boring becomes a
superordinate construct in the car construct system. To
ladder down to subordinate constructs the interviewer could
say "You mentioned that two of these cars are sports cars
and one is a saloon; tell me more about that?" The
respondent may reply "Well elements one and three are highly
streamlined while number two is shaped like a brick."
Streamlined---brick shaped becomes a subordinate construct
in the car construct system.
65
Using these sorts of questions the initial construct becomes
a basis from which other constructs are generated. Each new
construct casts more light on the way the respondent
construes the subject being researched and provides
additional background information and examples. Laddering
up the construct hierarchy tends to generate constructs more
personal and more related to the respondent's philosophy of
life. Laddering down generates more detailed and technical
details about the elements themselves.
At the end of the interview the interviewer will have
recorded a large number of polar constructs (usually between
ten and sixty) all of which provide insights on the subject
matter. A series of interviews with, for example, twenty
managers will typically yield between three and four hundred
constructs, although not all will be unique. These
constructs comprise, in themselves, a rich source of data.
iVa Data Analysis
The process can be terminated at this point, with the
respondent having benefited simplY by having their
constructs elicited and clarified. The interview
transcripts can also be content analysed if further
information is needed by the researcher. Alternatively the
respondent may be asked to rate the elements by their
constructs in terms of their unique grid or matrix. The
66
grid lists the elements used along the top and the
constructs elicited from the respondent down the side. The
respondent is invited to rate the extent to which each
construct applies to to each element. The grid can then be
used to explore the relationships between elements more
fully and can be analysed using one of a number of computer
packages that have been developed (for example Bell 1987).
An example (from Easterby-smith 1980) of a grid matrix is
shown in Figure 3:1. As can be seen in this example, the
elements are people with whom the respondent has a
relationship. The constructs are those that have been
obtained in an interview process similar to that outlined
above. The ticks and crosses mark the pole of the construct
most of the person used as the element. The
grid provides the respondent with a simple but useful
picture of similarities and differences in the people they
interact with. More sophisticated grids require responses
to rating scales rather than the simple binary approach
shown in Figure 3:1. This later approach permits more
sophisticated forms of statistical analysis.
67
FIGURE 3:1
ELEMENTS
1 2 3
( /) "Myself IIBoss" "Wife"
.
A Driving
-/
./
-I
B Mobile ~ X
v'
C Rigid X :; X
.
D Intelectual /
./
/
E critical X X X
4
"Best
Friend
ll
.
/
.
/
X
X
\/
5
Ii Person
Disliked)
X
X
.
-/
X
/
(X)
Easy-going
static
Open
Non-
intellectual
Accepting
(A dot (.) in the top left hand corner of a square indicates that the element above was one of the
'Itriad
il
that produced the construct for that row).
v. Comparing Construct Systems
Eliciting constructs from individuals through interviews
(with or without the extension to the grid) is a time
consuming process that may not be feasible in dealing with
large numbers of sUbjects. This was an issue in this
research as we wanted to use the R.G.T on a group large
enough to permit some generalizability in the findings. We
also wanted to compare our results across respondents,
particularly those at different organisational levels. An
individually formed and completed grid like that shown in
Figure 3:1 cannot by its nature, be compared with the grid
of another person. To make comparisons across respondents
requires a common grid which is completed by all
respondents.
One way of approaching this issue was to take a sample of
subjects and elicit a consensus of constructs from them as a
group. These constructs can in turn be used in a
standardised grid which is administered to larger numbers of
subjects (Crump et al 1981, Eden, Jones and simms 1983,
stewart and stewart 1981a, Ginsberg 1989) .. The interviews
generated a very large number of constructs, as the
respondents thought about managerial work. We felt that
reducing such a large number of constructs to the fifteen or
so constructs that might be included on a common grid was
making too great an imposition on the data. There were also
68
practical in arranging group sessions to
generate the original constructs. For these reasons this
process was not adopted.
Consequently we adopted the stewart and stewart
(1981a,1981b) approach of eliciting constructs from a
representative sample of a larger subject group and using
them as the basis for a questionnaire. The questionnaire
can then be administered to the larger group. The
questionnaire operates as a large grid, in this case with
many constructs (170 in the present study) and only two
elements. The respondents in the questionnaire survey rated
the two elements (most effective manager and least effective
manager) separately on each of the 170 constructs.
The questionnaire is comprised of categories generated by,
and in the language -of, the subjects themselves. The large
number of constructs used ensures that the full range of
constructs generated in the interviews are represented on
the questionnaire. It therefore retains the observer
neutralising characteristics of the interviews. As long as
it presents a representative range of constructs the
questionnaire can simultaneously tap the unique insights of
individuals and provide more generalizable findings. This
is the process used in this study. The interview approach,
the questionnaire formation process and the questionnaire
structure are explained in detail in chapter four.
69
CONCLUSION
The Repertory Grid Technique provides a means of eliciting
constructs of managerial effectiveness that reflect the
realities of individual and organisation settings, rather
than those of the researcher or research method. Both as a
technique and in its underlying assumptions, it is well
suited to the varied, complex and interactive nature of
managerial work. It proved to be a highly effective tool in
addressing the research question guiding this study. Its
application in this study is detailed in chapter four.
Subsequent chapters outline the results of this application
and their implications.
70
CHAPTER FOUR
DATA GATHERING
INTRODUCTION
Chapter two highlighted the lack of research on
effectiveness in management. The literature reviewed in
chapter two also indicated that there is very little
research eXRloring variations in managerial effectiveness
(in terms of characteristics and behaviour) between
different managerial levels. with these issues in mind we
developed the following research question; "What are the
characteristics and behaviours of effective versus
ineffective managers and how do these characteristics -and
behaviours vary between different managerial levels
lD
This question is the central focus of this study. In
addressing the research question, the data gathering process
proceeded through two phases. First, repertory grid
interviews (eighty-nine in total) were conducted with
managers and non-managers in a large public sector
organisation. These interviews were designed to elicit
constructs differentiating effective from ineffective
managers in the organisation. Second, a questionnaire study
was designed, in which the interview constructs were
introduced to a l a r g ~ r sample of managers and non-managers
71
in the organisation. As suggested in chapter three, the
questionnaire was developed from the constructs elicited in
the ipterviews. Three hundred and sixty-five questionnaire
responses were obtained.
In both the interview and questionnaire phases of the study,
criteria of managerial effectiveness were used as a base for
the generation of effectiveness categories and for
subsequent analysis. In this chapter we outline the data
gathering methods and the effectiveness criteria employed.
Issues relating to sampling and the generalizability of
results are also discussed. Specific modes of data analysis
are outlined in the chapters to which they are related. An
overview of the data analysis process is also provided at
the end of this chapter, as a guide to the reading of
subsequent chapters.
ORGANISATIONAL SETTING AND PARTICIPANTS
The study was conducted in the Department of Social Welfare,
with the data being collected during 1986 and 1987. This is
a large New Zealand public sector organisation which, at the
time of data gathering, employed slightly less than six
thousand staff (5943). The Department of Social Welfare was
at the time, undergoing a transformation from a stable
bureaucracy to a more innovative and accountable public
sector organisation. This study provided an opportunity to
72
assist the Department's managers in this transition and
concurrently gather data from which the research question
could be addressed.
73
Both the interview and questionnaire phases of the study
were conducted in four offices of the Department located in
Manukau, Hamilton, Nelson and Christchurch. Two of these
offices (Manukau and Christchurch) were large by Department
standards with 108 and 138 employees respectively. The
other two (Hamilton and Nelson) were medium in size with 65
and 69 staff respectively. An additional sixty
questionnaires were sent by mail to offices ina wide
variety of locations throughout New Zealand. In order to
ensure a good cross-sectional sample, a number of these mail
surveys went to small offices located in rural towns.
Both managerial and non-managerial respondents were sampled
in the study. The respondents were divided into four
management levels. Those below the supervisory Divisional
Officer level (104 grade level) were referred to as non-
managerial. The 104 grade Divisional Officer respondents
were classified as supervisory management. These managers
form the first line of supervisory management. Management
positions between the Divisional Officers and up to and
including the Assistant Directors, were classified as middle
management. District and Regional Directors were classified
as senior managers. The non-managerial, supervisory, middle
and senior management titles are used hereafter. The
management level criteria used to categorize the respondents
are in harmony with the literature in this area. The
literature and the classification of respondents into
management levels are discussed further in chapter seven.
All of the major, work areas were represented. These
included Benefits and Pensions, Administration, Social Work,
National Superannuation, Typing and District/Regional
Directors. A full outline of the interview and
questionnaire respondents is provided below.
SAMPLE DEFINITION
In chapter two we defined the manager as a person, usually
titled. manager who has responsibility, authority and
accountability for a discrete organisational unit and who
has authority and accountability for getting some of the
unit's work done through other people. This implies that
studies of managers should address themselves to people so
titled, who hold clear unit and staff management
responsibilities
In this study our concern was to explore the characteristics
and behaviours of individuals who met the above definition
of the term manager, particularly in terms of staff
management responsibility. To this end the specifically
technical, non-managerial sections of the organisation were
74
avoided (e.g. legal, planning etc). The focus was on units
characterised by staff/manager reporting relationships. As
we shall see below, the interview respondents recorded the
names of the managers they were comparing in the interview
study. A review of the names of these managers indicates
that all of them had staff management responsibility.
The questionnaire survey was more difficult to monitor, in
that the names of ratee managers were not indicated. The
respondents were asked however, to record the mapagement
level and title of the ratees on the questionnaire. A
review of the ratee managers identified in the questionnaire
study indicated that seventy-eight percent of the ratees
were either 104 level supervisors or Regional/District
Directors. All of these positions involve staff management.
The remaining ratees were senior Divisional Officers, Senior
Executive Officers, Assistant Directors and Area Welfare
Officers. These positions also involve staff management.
It appears that the great majority, if not all, of our
respondents would have identified the characteristics and
behaviours of managers who conformed to the management
definition outlined above. To this extent we can be
confident that the characteristics and behaviours reported
in this study are representative of managers in the
Department rather than of senior staff without managerial
responsibilities.
GENERALIZABILITY OF FINDINGS
The primary aim of this study was to e x p l ~ r e and model the
characteristics and behaviours of effective versus
ineffective managers. Because the study was undertaken in a
specific organisational setting it is important to consider
the extent to which the results can be generalized to
managers in the Department of Social Welfare as a whole and
to organisations beyond the Department.
As is discussed below, the sample percentages and response
rates are such that we may be confident that the
questionnaire results are representative of the Manukau,
Hamilton, Nelson and Christchurch offices (eighty-three
percent of the staff in these four offices were sampled in
the questionnaire study). There are also indications that
the findings may be generalizable to management in the
Department as a whole and to organisations outside of the
Department.
The size of the Department and the geographic dispersal of
its staff and offices prevented the development of a simple
random sample (Tull and Hawkins 1976, p.159). As an
alternative, a representative judgment sample (Tull and
Hawkins 1976, p.161) was developed. This sample was deemed
representative of the organisation, by management
representatives and external consultants who had an intimate
76
and long term association with the Department. As is
outlined above, th,e sample included a variety of offices,
(large, medium and small in urban and rural settings) work
types, management levels and both sexes. The sample
appeared to be at least as representative of the total
organisation as a strictly random sample would have been.
Total employee numbers, together with the interview and
questionnaire responses at each management level, are shown
in Table 4:1. These figures allow us to make an estimation
of the representativeness of the interview and questionnaire
data. Application of standard formulae (Tull and Hawkins
1976, Gimpl 1990) for determining appropriate sample size
(for random samples) indicates that the 365 (6.14 percent of
total staff) questionnaire responses would be representative
(at the 95% confidence level) of the total Department
between +/-3 to 4%, depending on the amount of agreement in
responses to a given question. Use of the same formulae
indicates lower levels of error in the responses of the
managerial population of which 11.87 percent were sampled
(see Table 4:1). As discussed above, the sample appears to
be at least as representative as a strictly random sample.
Given this assumption, the sample size estimates that
emerged in applying standard formulae for random samples,
provide reasonable confidence that the questionnaire
responses are representative of the Department as a whole.
77
TABLE 4': 1
RESPONDENT SAMPLE
Emgloyees Interview Samgle Questionnaire
Samgle
No. Percent No. Percent
Senior Management 83 8 (9.64) 28 (33.73)
Middle Management 662 41 (6.19) 75 (11.33)
Supervisory 999 15 (1. 5) 104 (10.41)
Management
Total Managers (3.67) (11.87)
Non-Management 4199 25 (.0. 59 ) 158 (3.76)
Total Employees (1.50) (6.14)
The smaller sample size (with the exception of the middle
and senior management groups) of the interview study places
greater limitations on its generalizability to the
Department as a whole. As is discussed below, the more
important issue is the extent to which the constructs
elicited in the interviews presented questionnaire
respondents with a range of constructs or ideas
representative of their work experience; that is, the extent
to which the questionnaire was content valid. This aim
appears to have been accomplished.
78
There are strong indications that the findings of this study
can be generalized to organisations beyond the Department of
Social Welfare. Two pieces of evidence support this
conclusion. The first relates to other studies of
managerial effectiveness conducted (by the author and
colleagues) in New Zealand ( L ~ m 1984, Dakin and Cammock
1985). One of these "studies was conducted in the private
sector with two further studies being conducted in
additional public sector organisations. All of these
studies used the Repertory Grid Technique. Both the
constructs and construct categories that emerged from these
studies show a remarkable similarity, both to each other and
to those generated in this study. Given the premise of
diversity underlying this study and the capacity of the
Repertory Grid Technique to tap into such diversity we found
the similarity in these findings of great interest.
The second piece of evidence relates to the similarity
between the results of this study and studies of managerial
work conducted overseas. In chapter six we outline the
results of a factor analysis of the questionnaire data,
conducted as part of this study. It is of interest that the
factor structure emerging from this study is in direct
conformity with the agenda building and networking
categories reported by Kotter (1982, 1988). The parallels
are striking given the variations in both research settings
and methods, between this and Kotter's work. The factor
model also echoes the widely recognised skill typology of
Katz (1974). These similarities indicate that the findings
of this research may well have a generalizability beyond
this particular research setting. They support the limited
generalizations which are drawn from the results described
in later chapters. They also lend support to the
possibility of generic models of effective managing which
have applicability (although with quantitative variations in
emphasis) across different management settings. This
conclusion and the need for further research in exploration
of this issue is discussed further in chapter eight.
79
INTERVIEW :STUDY
Interview iBample
Having established a research setting we next had to define
the perceived characteristics and behaviours of effective
and ineffective managers in t.he organisation. As discussed
in chapters two and three, the Repertory Grid Technique
(R.G.T) was used to elicit constructs about managerial
effectiveness from respondents representative of the
Department of Social Welfare.
Eighty-nine'Repertory Grid interviews were conducted in the
Christchurch, Hamilton, and Manakau offices of the
Department. Eighty-eight of the interviews provided usable
responses. In keeping with t:he method described in chapter
three, the constructs generat:ed in the interviews were to be
used to develop a questionnaire for distribution to a larger
sample of Department Clearly the questionnaire
itself had to be representative of the work experience of
this wider sample. As a consequence the representativeness
of the interview sample was a critical issue.
This iSSUE! was addressed by !selecting (with the assistance
of the management and consultancy group mentioned above) an
interview sample which was as representative as possible (in
terms of, geographical area, organisational level, and
80
81
respondent sex) of the Department. The interview sample is
shown in Table 4.2. The Manukau and Christchurch offices
are representative of larger urban areas, while Hamilton and
Nelson represent smaller urban/rural populations. Some
sampling compromises were made, dictated by the
practicalities of interview time, travel cost and
accessibility of respondents. Ideally, the sample would
have contained more female respondents, more respondents
from the Nelson office and relatively fewer middle managers.
However, the larger numbers of middle management respondents
ensured a good representation of senior, middle and
supervisory management ratees.
Overall, there is good reason to believe that the interview
constructs are representative of the total organisation. On
reviewing the interview transcripts it is clear ~ h a t there
is a great deal ~ f overlap in the constructs generated by
different respondents. All of the constructs were mentioned
by at least two respondents, with the great majority being
mentioned by a large number of respondents. Because of this
overlap a smaller number of respondents would probably have
generated a very similar construct sample. The larger
numbers of interviews however, adds somewhat to the
diversity of constructs and provides a guide to the emphasis
placed on different categories. Additional interviews were
also beneficial in reflecting the different perspectives of
TABLE 4:2
INTERVIEW RESPONDENTS BY OFFICE, LEVEL AND SEX
Respondent Office Respondent Level Respondent Sex
Manukua 14 Senior 7 Male 46
Hamilton = 28 Middle 43 Female 40
Nelson = 3 Supervisory 13 Unknown 2
Christchurch
!.l
Non-Management 25
88 88 88
respondents at different organisational levels (see chapter
seven) .
Interview Format
The interview study was designed to elicit constructs
differentiating effective from ineffective managers in the
Department. The interviews were conducted by the writer and
(on a few occassions) two trained graduate assistants. They
followed the format outlined in chapter three. As we saw in
chapter three, the choice of elements being used is dictated
by the nature of the domain being studied. In the case of
the present research, because we were interested in
manageri?l effectiveness, we invited respondents to -think
about most and least effective managers as the element set.
The element set used in the interviews is shown in Table
4:3.
other types of elements could have been used. stewart and
stewart (1981a, 1981b) for example, report the use of work
events as elements. stewart and stewart (1981b) observed
that the use of people as elements also promotes effective
construct elicitation. Our pre-testing confirmed stewart
and stewart's (1981b) observation. Pre-testing showed the
use of other people (i.e. other managers) as elements to be
more successful than the use of managerial events. The
manager element set shown in Table 4:3 worked extremely
82
TABLE 4: 3
INTERVIEW ELEMENTS
1. Your most effective peer manager.
2. Yourself.
3. Your least effective peer manager.
4.. Your most effective subordinate manager.
5. Your least effective subordinate manager.
6. Another subordinate manager who is highly effective.
7. Your boss at the next level.
8. Apart from (7) the least effective manager you know at that level.
9. Apart from (7) the most effective manager you know at that level.
well, generating high levels of interest, involvement and
candour on the part of the
Because of our interest in hierarchical differences in
perceptions of effectiveness and to reflect the varied
perspectives of peers, superiors and subordinates (see
Perry and Mahoney 1956, Gordon and Medland 1965,
Schwartz, Stark and Schiffman 1970) respondents were asked
to think about managers at their own level and at the levels
immediately above and below them. However, non-supervisors
(who had no peer or subordinate managers) considered only
the level above them (i.e. supervisors). First line
supervisors considered only their own level and the level
immediately above them. The level of focus requested of the
interview respondents is shown in Table 4:4.
Interview Process
After a brief outline of the purpose of the interview and an
assurance of confidentiality, the respondents were given
three blank cards. On these cards they wrote the names of
the three managers selected as elements one, two and three.
On card one they wrote the name of the most effective
manager they knew at their own level. On card two they
wrote their own name and on card three they wrote the name
of the least effective manager they knew at their own level.
Three new cards were given as each new element set was
83
..-l
QJ
:::-
QJ
H
QJ
QJ
+J
rd
0:::
Subordinate
Managers
(Elements
4, 5, 6)
Peer Managers
(Elements
1, 2, 3)
Superior
Managers
(Elements
7, 8, 9)
TABLE 4:4
LEVEL OF FOCUS REQUESTED OF INTERVIEW RESPONDENTS
Non-Management
Respondents
/
Respondent Level
Supervisory
Respondents
vi
~
Middle
Management
Respondents
vi
\/
/
Senior
Management
Respondents
\/
/
V
\/
introduced. The nine elements were presented to the
respondents in three triad sets. The following pattern was
used.
Triad Set Combination
1 1, 2, 3
4, 5, 6
3 7, 8, 9
This grouping ensured that comparisons were always made
between managers at the same level, thus avoiding any cross-
level confounding. It should be noted that, in comparing
managers, no attempts were made to get respondents to make
comparisons across managerial levels. Cross-level
comparisons were avoided in order that the constructs
elicited could be uniquely cross-referenced to particular
managerial levels.
with the element names recorded on the cards the respondents
were asked "Can you tell me one way in which any two of
these managers are similar but different from the third?"
This question would elicit an initial construct. Once the
initial construct was elicited construct relationships were
further explored using Hinkle's (1965, 1970) laddering
84
technique. The question "Why is that important?" was used
to ladder-up the construct hierarchy. The questions "Why is
that?" "What are the implications of that?" and "Can you
tell me more about that?" were used to ladder-down the
construct hierarchy.
As an example of the interview process, we will look at part
of an interview conducted with a Regional Director (senior
manager). The interview had progressed through the first
three elements and was now approaching the second triad
(elements four, five and six). We pick up the process as
the interviewer is introducing elements four, five and six.
The interviewer commences "I would like you to now consider
three more managers, this time at the level below you.
First on this card, (number four) I would like you to write
the name of your most effective subordinate manager." The
respondent wrote the name Jill Todd (all of the names are
disguised). "On this next card I would like you to write
the name of your least effective subordinate manager." The
respondent wrote the name Joy Mullens. "Finally I would
like you to write the name of another subordinate manager
who is highly effective." The respondent wrote the name
Roger Bright.
85
At the end of this process the respondent had three separate
cards in front of him, with the names of three different
managers who formed elements four, five and six. The
respondent was asked to lay the cards out in front of him
and think about the named individuals, as managers. After a
suitable period for reflection he was asked "Can you tell
me one way in which two of these managers are similar but
different from the third?" The manager replied. "Yes, four
and six use consultation, whereas number five is very
secretive and unable to share information."
The interviewer then wrote down the construct; Uses
Consultation-------- Very secretive, unable to share
information. To further explore this construct the
respondent was asked the laddering question "Why is that?"
The following r ~ s p o n s e emerged which provided a subordinate
construct and further illuminated the consultation issue.
"Well, four and six are more secure in their position, they
don't feel like they have to defend their right to be in
charge. By contrast, number five is less secure and as a
result she's always defending her right to be in charge".
This response was recorded as the construct; Secure in the
position and don't feel they have to defend their right to
be in charge-------- Less secure in the position, is always
defending her right to be in charge.
To explore the issue further the question "Why is that?" was
again asked. The respondent replied "Four and six have an
acknowledged level of technical expertise and they feel
secure in that knowledge. Five has less technical knowledge
86
and consequently is less secure in her role". The
interviewer recorded the construct; Acknowledged level of
technical expertise-------Less technical knowledge. ~ h r o u g h
the interview process two subordinate constructs were
developed from the original.
At the start of this part of the interview the respondent
distinguished ratee managers by the level of consultation
used in their managerial approaches. Through the laddering
approach the interview explored the underlying reasons for
the use or non-use of consultation. We discovered that the
use of consultation relates to the manager's level of
confidence and security in the position, which in turn is
related to levels of technical knowledge.
When the initial construct was fully developed the
respondent was then asked "Can you tell me any other way in
which two of these managers are similar, but different from
the third?'! This question was repeated until the range of
differences was fully explored. The process was then
repeated with the elements seven, eight and nine. This
process was repeated to the point where no new constructs
emerged and the interview was concluded.
87
In all cases, managers labelled as effective were contrasted
with those labelled ineffective. As a result the interview
responses emerged as as series of bi-polar descriptions
(constructs) of the characteristics and behaviours of most
versus least effective managers. One side was descriptive
of the characteristics and behaviours of managers perceived
as most effective the other of managers perceived as least
effective. As mentioned in previous chapters, Appendix One
shows a page of recorded interview responses.
As outlined above, respondents were asked to record the
names of the managers identified (elements) on the cards.
These cards were retained at the end of the interview to
ensure that the constructs were aligned with the correct
management level in the analysis. They were also used as a
check on the reliability of respondent assessments of
managerial effectiveness, and to ensure that the ratee
managers held staff management responsibilities (see above) .
88
The interviews took between one and two hours to complete
and almost invariably generated a high level of respondent
interest and involvement. A number of more senior
respondents found the process useful in considering their
own approach to management and in reviewing the activity of
their subordinates. Each interview generated a number of
constructs which described the perceived differences between
',effective and effective managers in the Department. The
interviews also provided a wealth of examples and anecdotes
which expanded and explained the constructs. The
constructs developed in the interviews, were used in two
ways. First, they were content analysed to identify
differences in the frequencies with which constructs were
used by managers at different organisational levels.
Second, they provided the items used in the subsequent
questionnaire study.
QUESTIONNAIRE STUDY
Questionnaire Rationale
Repertory Grid interviews provide a tremendous depth of
understanding of individual perspectives. The interviews
share the limitations of observation studies however, in
that the time involved makes it difficult to apply the
technique to large respondent samples. In chapter three we
discussed various alternatives that can be pursued to extend
the teChnique to larger samples of respondents. The
approach employed in this study followed the prescription of
Stewart and Stewart (1981a, 1981b) in that it used the
interview constructs to form a questionnaire, which was then
applied to a larger subject group.
Questionnaire Development
The following procedure was used in developing the
questionnaire;
89
1. The interview constructs were transcribed from the
interview protocols onto index cards. Respondents
frequently used the same, or very similar constructs. Only
unique constructs were recorded in the transcription
process. The total number of unique constructs transcribed
was around three hundred.
2. The constructs were sorted into conceptually similar
categories using a modified Q-Sort technique (Guilford 1954,
Tull and Hawkins 1976). six judges worked independently to
sort the constructs into conceptually similar categories.
The judges then met and were able to agree on a core of
twenty-one logical categories, without difficulty.
3. The categories were then used as a guide to the
inclusion of in the questionnaire. The judges
agreed on a core of 170 constructs which were used as items
in the finalised questionnaire. The remaining constructs
were dropped because of their similarity to those retained.
For example the constructs: Separates his work and private
life -----Mixes his work and private life, and Does not
allow his work and private life to interfere with one
another ----- Allows his work and private life to interfere
with one another, clearly address the same issues. Only the
latter construct was retained in the questionnaire. The
finalised questionnaire is shown as Appendix Two.
90
It may seem that the use of judges to classify constructs is
antithetical to the spirit of Personal Construct Theory,
which seeks to explore the unique worlds of individual
respondents. However, as we have noted above, the study
also seeks generality, and as such needed to use a standard
grid across a number of respondents. Further, although
there is some danger of losing unique perspectives in such
an editing process, the task of the judges proved to be
quite easy. Constructs which were similar were quite easy
to spot and it was a straightforward task to eliminate
constructs which were redundant. Constructs which were'
clearly unique were also relatively easy to identify and
include.
Overall, we were satisfied that the final questionnaire was
representative of the range of perspectives held by
individuals in the Department. It was therefore felt
unlikely to suffer from the sort of deficiency problems
outlined in chapter two. The questionnaire, of course, was
not as specifically tailored to individual perspectives as
365 repertory interviews might have been. Nevertheless, it
represented a useful compromise that permitted a greater
sample size and still allowed individual respondents to
express themselves along relevant dimensions.
As discussed below, respondents were asked to complete the
questionnaire twice, on one occasion rating the most
91
effective manager they knew and then rating the least
effective manager they knew. In this way it was possible to
explore the perceived differences between managers believed
to be effective and ineffective. By presenting constructs
representative of those used by the respondents, the
questionnaire retained the benefits of the Repertory Grid
Technique while providing more generalizable data.
Questionnaire Administration
The questionnaires contained two identical sections, each
with 170 items (see Appendix Two). On one section the
respondents were requested to rate a most effective manager.
On the other they were requested to rate a least effective
manager. As can be seen from Appendix Two the questionnaire
items are bi-polar (for example, Poor Listener: discourages
discussion ----- Listens well: encourages discussion). The
respondents were requested to place a tick on a five point
scale indicating the extent to which the item descriptors
were descriptive of the ratee. To avoid response sets the
item polarity was reversed every five items.
The order in which respondents answered the questionnaire
was randomly decided. Approximately half the sample rated a
most effective manager first, followed by a least effective
manager. The other half of the sample received
questionnaires which reversed the order. Respondents were
92
asked to leave at least a few hours and if possible a full
day between the most and least effective ratings. The
reversal of items and the gap b e t w ~ e n most and least
effective manager ratings appears to have limited the impact
of halo effects in the responses (see chapters five and
six) .
93
Because the questionnaire was to be administered at
different hierarchical levels, and to control for the level
of the managers being rated, three versions of the
questionnaire were produced. Version one requested
respondents to rate 104 level supervisory managers.
Version two requested respondents to rate peer managers.
Version three requested respondents to rate superior
managers. Version four requested respondents to rate
subordinate managers. Non-managerial respondents rated only
their immediate supervisors and used version one.
Supervisors rated peers and their immediate superiors and
used versions two and three. Middle and senior management
respondents rated either subordinate, peer or superior
managers and used versions two, three and four. Table 4:5
shows the pattern of questionnaire distribution.
The ratee category (and questionnaire version) administered
to the respondents, was randomly determined. The exception
to this was the mail survey in which the administration was
manipulated to ensure that the respondents rated only
TABLE 4:5
QUESTIONNAIRES
Version One Version Two Version Three Version Four
(Rating (Rating Peers) (Rating Superiors) (Rating
Supervisors) Subordinates)
Respondents
/
Non-Managerial
V
----- ----- -----
Supervisors -----
\/ \/
-----
Middle-Managers -----
\/ \/ \/
Senior Managers -----
\/ "\/ \/
District or Regional Directors. This approach was used to
increase the number of ratees at the senior level.
Respondents were requested to write the grade and job title
(but not the name) of the person they were rating to ensure
the questionnaire analysis was correctly focused.
Questionnaire Sample
The questionnaires were distributed directly to staff, in
the Department of Social Welfare, in the Christchurch,
Nelson, Hamilton and Manakau offices. At each of these
offices an explanatory meeting was held, prior to the
distribution of the questionnaires. At this meeting
questions and concerns were addressed. The instructions on
the front of questionnaires were and any areas
of confusion clarified. As discussed above, a mail survey
was also conducted covering an additional 60 respondents.
An explanatory letter was included with the mail survey
questionnaires.
The office respondents were visited individually on two
further occasions. First to answer any queries and second
to pick up the completed questionnaires. with the exception
of the Hamilton office an attempt was made to survey all of
the staff at each office. The work commitments of the
Hamilton office were such that not all of the staff were
able to take part. In this office all of the available
94
senior staff and fifty percent of the remaining staff were
sampled. A follow-up phone call was made to mail survey
respondents who failed to respond within four weeks. The
close follow-up yielded a high response rate from both
office and mail survey respondents. The response rates for
the office and mail survey respondents are shown in table
95
4:6.
Usable questionnaire responses were received from 365
respondents. The questionnaire response patterns for the
mail survey and each office, work area, management level and
respondent sex are shown in Table 4:7. As can be seen from
Table 4:7, the questionnaire respondents represent a range
of offices, work areas, management levels and both sexes.
As is discussed above, the size and diversity of the sample
gave reasonable confidence that the questionnaire results
were representative of the organisation as a whole.
MEASURING EFFECTIVENESS
Effectiveness criteria
Having developed methods for eliciting managerial
characteristics and behaviours the next requirement was to
ground these characteristics and behaviours against measures
of effectiveness. As we saw in chapter two, a variety of
measurement approaches have been employed in past research.
TABLE 4:6
RESPONSE RATES BY OFFICE
Office No. of Emgloyees Usable Resgonses Resgonse
Percentage
Manukau 108 104 96%
Hamilton 65 35 54%
Nelson 69 53 77%
Christchurch 138 135 98%
Mail 60 38 63%
Total 440 365 83%
TABLE 4:7
QUESTIONNAIRE RESPONDENTS BY OFFICE, WORK AREA, MANAGEMENT LEVEL, SEX
Office Work Area Manaqement Level Sex
Manukau 104 Benefits & Pensions 172 Senior Management 28 Male 110
Hamilton 35 Administration 70 Middle Management 75 Female 191
Nelson 53 Social Work 42 Supervisory Management 104 Unknown 64
Christchurch 135 National 39 Non-Management 158
Superannuation
Mail 38 Typing 14
District/Regional 28
Directors
365 365 365 365
While the preference in this study was to employ multiple
effectiveness criteria, practical limitations meant that
this was not possible in the Department of Social Welfare.
As a welfare organisation, the Department employed none of
the economic outcome measures often used in effectiveness
studies (for example profitability and return on
investment). other objective effectiveness criteria which
might have proved useful, such as budgeted versus actual
costs, efficiency of resource usage and staff morale, were
not sufficiently developed to be useful in this study.
The performance appraisal and promotion systems were highly
formalised and documentation existed in this area on all of
the managers in the organisation. This information however,
was regarded as too sensitive to be made available for use
in the study. Apart from this, both the appraisal and
promotional systems were seen by many staff as inadequate
and inequitable. At the time of the study, a major review
of both systems was in progress. Failure to get access to
promotional and appraisal information and problems within
these systems ruled out their use in this study. It was not
possible therefore, to employ the sort of effectiveness
indexes and global rankings described in chapter two (see
Morse and Wagner 1978, Luthans et al 1985, Martinko and
Gardiner 1990)
96
97
In chapter two we discussed the preference expressed by
Hales (1986) for more contingent effectiveness measures.
Hales (1986, p.l08) writes "one such contingent standard
with which to compare actual managerial practice might be
what others expect or require managers to do. Good or bad
performance may then be conceived in terms of the extent to
which managers' performance matches others' expectations".
The effectiveness measure employed in this study followed
Hale's prescription. In the interview study the elements
referred to most effective and lea-st effective managers.
The questionnaire study followed the same approach, inviting
respondents to rate a most effective and a least effective
manager. This classification was reinforced, at the end of
each questionnaire section, by a five point global
. .
effectiveness rating (see questionnaire Appendix Two) .
Respondents were invited to rate each ratee manager on this
five point scale (ranging from Below Average/Bottom 10% to
Superior/Top 10%). As discussed in chapter two, previous
research indicates that a single broad subjective measure
(such as the most effective/least effective designation used
here) is probably as useful in measuring effectiveness as
are "objective" criteria or a battery of subjective ratings
(Jaques 1976, smith 1976, Hiles 1986, Barker, Tjosvold and
Andrews 1988, Nathan and Alexander 1988). While we would
have liked to have employed multiple effectiveness criteria
the evidence suggests that little has been lost with the
broad subjective criteria used in the study.
98
The sUbjective criteria used in this study echoed the
approach employed by Flanagan (1951, 1952) and that of Roach
(1956). More recently Barker et al (1988) used the same
criteria in a study of the conflict approaches of project
managers. Respondents in the Barker et al (1988) study were
requested to describe the behaviour of the most effective
and the least effective project managers they had ever
worked with. The conflict resolution approaches of most and
least effective managers were then compared. The study of
Barker et al (1988) also employed a more comprehensive
battery of subjective effectiveness criteria. For example,
the impact of project managers behaviour on the job
satisfaction and commitment of staff. These additional
criteria produced results identical to the broad most and
least effective designation.
In the present study, several of the analyses were run
including only most effective managers with global ratings
of four and five (very good and superior) and least
effective managers with global ratings of one and two (below
average and average). In all cases no significant
difference was found between questionnaire analyses using
the high and low global effectiveness ratings and those
using the broad, most effective/least effective designation.
The analysis presented in subsequent chapters therefore,
employs only the broad most effective/least effective
criterion.
The Reliability of the Effectiveness criteria
During the interview phase of the study, respondents were
asked to nominate managers who they saw as most and least
effective. Sixty-eight names were nominated by the
interview respondents, some classified as most effective,
others as least effective. All of the sixty-eight nominees
were identified by more than one respondent. In fifty-one
of the sixty-eight cases there was 100% agreement between
respondents about whether a person was effective or
ineffective. That is; fifty-one of the sixty-eight
nominated managers were designated either most or least
effective by all of the respondents who identified them.
There was minor disagreement over the remaining seventeen
nominees, but overall, agreement in assigning people to most
and least effective manager categories was 90.6%. This
indicates a high level of reliability in the judgements of
the respondents and indicates a correspondingly high level
of reliability in the effectiveness criteria employed in
this study.
99
CONCLUSION AND OVERVIEW OF DATA ANALYSIS
In this chi:tpter we have outlined the methods used in
developing characteristic and behavioural categories
descriptivl:! of most and least effective management in the
Department of Social Welfare. We have also discussed the
criteria u:3ed in defining effective and ineffective
managers. These criteria provided the grounding against
which the I::haracteristic and behavioural categories were
developed. Issues relating to sampling and the
generalizability of the research "findings have also been
discussed.
The data gathering phase of the research produced a wealth
of data. 'rhe specific techniques used in analysing this
data are d,etailed in subsequent chapters. The following
overview of the data analysis is provided as a guide to
further re,:lding.
With the field work completed. the first step was to analyse
the questi'onnaire results. F'rom this initial analysis
dependent variables were identified which formed the
backbone of subsequent questi.onnaire and interview analyses.
The questionnaire responses were used to cluster the 170
questionnaire items into twenty categories (hereafter
referred to as scales) of tWCI to twelve items. The scales
described the characteristics; and behaviours of the most and
100
least effective ratee managers. The scales and the
clustering technique are described in chapter five.
The scales were in turn factor analysed to a two factor
solution. In addition to the two factors, the scale
technical knowledge emerged as a distinct third dimension.
The factor analysis, the resultant factor structure and the
implications of the structure are discussed in chapter six.
The scale and factor categories were used as dependent
variables in testing hypotheses relating to variations in
the characteristics and behaviours of most/least effective
managers between managerial levels. The interview content
analysis, hypotheses, the procedures used in hypotheses
testing, the results and their implications are discussed in
chapter seven.
Chapter eight concludes this thesis. It highlights the
conclusions that emerge from the study as a whole and
discusses their implications for managerial research,
teaching arid development.
101
CHAPTER FIVE
DEFINING THE CHARACTERISTICS AND BEHAVIOURS OF MOST AND
LEAST EFFECTIVE MANAGERS
INTRODUCTION
As we have seen in previous chapters, the p r i n c i p ~ l
objective of this study has been to describe the
characteristics and behaviours of effective versus
ineffective managers. In chapter four we described the
questionnaire development process, which generated a
questionnaire with 170 items. The number of questionnaire
items was much too large to provide a succinct
characteristic or behavioural description. The need was to
reduce the 170 questionnaire items to a smaller number-of
categories. In this chapter we describe the reduction
process, which was achieved by forming the items into
logical categories (using six judges) and then using
Pearson's Correlation Coefficients and Cronbach's Alpha to
test and finalise the categories. This process generated
twenty scales descriptive of the characteristics and
behaviours of most and least effective managers in the
Department of Social Welfare. In this chapter we use these
scales to describe effective and ineffective management in
102
the Department. We explore differences in emphasis on the
scales, between ratings of most and least effective
managers, and compare them with previous research findings.
We also look at the issue of interaction and overlap between
the scales. The methods used, the findings and their
implications are discussed in this chapter. The results
outlined in this chapter have important implications for
managerial development. These implications are touched on
in this chapter and are discussed in greater depth in
chapter eight.
DATA ANALYSIS
Developing Characteristic and Behavioural categories
Item Reduction
As mentioned above, the principle objective of this study
was to define the characteristics and behaviours of
effective versus ineffective managers. Following the
interview study, 170 descriptions of managerial behaviour
and characteristics had been edited into a questionnaire
(see Chapter Four). Responses to the questionnaire were
used to reduce the items into a smaller set of scale
categories.
103
The ratio of subjects to variables (approximately 2 to 1)
precluded the use of data reduction by factor analysis or
similar clustering procedures. The recommended ratio for
reliable use of factor analysis is 10-20:1. Given the size
of the questionnaire (170 items), a sample of 2500 would
have been needed for the reliable use of factor analysis. A
sample of this magnitude would have exceeded both the number
of managers in the Department and the resources of this
study.
The item reduction process proceeded as follows;
104
step One
Only items which discriminated significantly between most
and least effective managers were included in the analysis.
That is, we were concerned to include in the analysis only
items which represented true differences in effectiveness.
T-tests (related samples) were conducted between all pairs
of items on alternative forms of the questionnaire (most
versus least effective). All of the items discriminated
between most and least effective managers at or above the
.0001 level. Item 19 discriminated in reverse of the
expected direction and was dropped from the analysis.
This result is consistent with the stewart's (1981a) finding
that use of the Repertory Grid Technique yields a high
proportion of constructs which discriminate in terms of
effectiveness. However, the very ,high proportion of
discriminating items (100%) a concern over the
influence of halo error. It is possible that
questionnaire respondents may have felt that most effective
managers should be rated high across all the items and the
least effective managers should be rated low. This could
account for the large and consistent differences between
most and least effective ratings. As noted in chapter four,
the polarity of the questionnaire items was reversed every
five items. The respondents were also requested to rate
specific individuals and to leave a time gap between most
and least effective ratings. These steps were taken
specifically to limit the impact of halo effects in the
questionnaire responses. There is strong evidence that this
approach' was successful and that halo error is not a
significant issue in the results of this This
evidence is discussed further below and in chapter six.
step Two
The process used to reduce the 170 questionnaire items
proceeded as follows;
105
1. Formation of Logical Scales
In outlining the questionnaire development process in
chapter four, we noted that six judges worked to sort the
interview constructs into twenty-one logical categories. In
this process approximately three-hundred constructs from the
Repertory Grid interviews were transcribed onto cards. The
six judges worked independently with the same construct set.
Their instructions were to sort the constructs into distinct
logical categories. The judges groupings showed-a strong
consensus. They then met to discuss their groupings and
remove redundant constructs. The judges arrived at an
agreed set of twenty-one construct groups (with 170
constructs) without difficulty. These construct groups were
used as the initial logical scale categories.
2. Checking Internal Scale consistency
Using the scales defined by the six judges, each scale (for
both most and least effective manager ratings) was examined
for internal consistency using Cronbach's Alpha. The SPSSx
Reliability programme was employed. Cronbach's Alpha is
defined as follows (see Norusis, 1988, p.207).
106
kr
1+(k-1) r
k is the number of items in the scale and r is the average
correlation between items in the scale. Hence alpha can
range in value from 0 to 1. Alpha increases in size, for a
constant r, as the number of scale items increases.
3. Reassignment of Items and Scale Finalisation
Item loadings on each scale were computed using Pearson's
Correlation Coefficients for all 170 items. Pearson's
Correlation Coefficients were used as g u i d ~ in reassigning
items that detracteq from the alpha scores of the scales
they were originally assigned to. These items were
reassigned to scales with which they had a significant
correlation and the reliability analysis re-conducted. In
almost all cases the items ended up in the scale with which
they were most highly correlated. On some occassions items
were assigned to a scale with which they did not have the
highest correlation but with which they appeared to be
logically related. In no case were items assigned to scales
with which they were not strongly correlated.
Thirteen items (9, 22, 30, 41, 46, 55, 56, 69, 83, 93, 119,
126, and 131) were dropped from the analysis at this point,
107
as they detracted from the alpha scores of all the scales
and did not form any separate and cohesive scales within
themselves. Along with item nineteen this took the total
number of items dropped from the analysis to fourteen. The
scales delegation and training were combined into one, as
their items correlated strongly with both scale categories.
At the end of the reliability analyses we had twenty scale
categories. All of the scales, with the exception of the
external networking scale, are robust in that their alphas
exceed the .50 level recommended as adequate for research
purposes (Nunally 1967). The external networking scale was
of concern as its internal consistency falls below the .50
alpha level. As a consequence it was not included in the
analyses described below, nor in the factor analysis
described in chapter six. It has been included however, in
the analysis of variance described in_chapter seven.
Exploring Patterns of Emphasis on the Scale categories
The twenty scales serve to define the characteristics and
behaviours of most and least effective managers in the
Department of Social Welfare. The next task was to explore
differences in the patterns of emphasis on these scales
between ratings of most and least effective managers. To
explore this issue scale mean scores and rankings were
calculated for both most and least effective questionnaire
responses. Multiple t-tests were used (for most and least
108
effective manager ratings) to test for significance in
differences between the mean scores and rankings of the
scale categories. Pearson's correlation coefficients were
used to further explore variation between most and least
effective manager ratings. Ratings of most and least
effective managers on each of the scale categories were
correlated.
RESULTS AND DISCUSSION
Describing The Characteristics and Behaviours of Most and
Least Managers
Using the reliability analysis discussed above, we have been
able to define twenty scales descriptive of
characteristics and behaviours of effective and ineffective
managers in the Department of Social Welfare. The finalised
scales, their items and reliability coefficients, are shown
in Table 5:1. As shown in Table 5:1, each characteristic
and behavioural category is comprised of two to thirteen
questionnaire items. It will be noted from this table, that
the reliability coefficients (alphas) are larger in all
cases but one (scale seventeen) for the least effective
manager scales.
109
TABLE S: 1
SCALE CATEGORIES (CHARACTERISTICS AND BEHAVIOURS)
Scale category Items Reliability Coefficients
CONCEPTUAL ABILITY (Most Effective) (Least Effective)
Managers Managers
I. Goal Setting 87,134,145 .7086 .7687
2. Innovation 5,10,67,88,106,108, .7688 .8688
110,114,116
3. Future Orientation 72,75,89,105 .6714 .7618
4. overview 51,65,68,73,74,78, .8116 .8471
91,127,142,154
5. Managing/operating 60,80,136,153 .6124 .6175
6. Stress Management 16,130 .5664 .748]
7. Hork Capacity 1,13,15,62,63,102, .7894 .8480
121,125,128,149,165
8. Assertiveness 3,7,33,49,70,85,95, .8443 .87]2
97,101,122,150,158,
9. Priori tis ing 71,76,104,107,111, .7917 .8187
112,115,124
10. Problem Solving 4,12,79,81,82,84,90, .8756 .8919
II. Personal 2,54,58,66,86,96,103 .7484 .8249
Organisation
INTERPERSONAL ABILITY
12. Delegation/ 26,27,34,38,40,44, .7766 .7909
Training 117,132,143,
13. Consultation 6,8,11,14,29,45,77, .8187 .8816
123,133,146,151,168
14. Feedback 18,28,31,118,120, .8439 .8788
138,139,152,156,167
15. Team Building 17,20,21,43,48,140, .879] .898]
157,160,164,169,170
16. Concern for Others 2],24,25,]2,52, .7910 .8671
129,141,144,161
17. Personality 35,37,50,53,59, .8096 .808]
92,94,100,162
18. Integrity 36,39,42,47,64,98, .7724 .8490
155,163
OTHERS
19. Technical Knowledge 57,61 .5480 .5878
20. External Networking 159, 166 .4800 .5522
A fuller description of the scale categories is provided in
Appendix Three. The descriptions provided in Appendix Three
are developed by bringing together the questionnaire items
included in each category. The most effective managers are
described first, followed by the least effective managers.
Each description presents an extreme picture, with most
managers lying somewhere between the two poles. Interview
quotations which are illustrative of the characteristics and
behaviours are outlined below each description.
As a pre-cursor to chapter six, the scales (both in Table
5:1 and Appendix Three) are grouped under three categories
labelled Conceptual ability, Interpersonal ability and
Others. The conceptual and interpersonal ability categories
emerged from the factor analysis (detailed in chapter six)
and are consistent with previous taxonomies, such as that of
Katz (1974) and Kotter (1982). Each of the twenty scales
are descriptive of specific aspects of managerial
effectiveness. Overall, the scales provide a direct
reflection of the realities of managerial effectiveness and
ineffectiveness in the Department of Social Welfare. They
have already provided a useful guide to management
development efforts in the Department and as discussed in
chapter four, may have relevance in other private and public
sector organisations. For the purposes of this study, we do
not wish to detail the specific developmental needs
suggested by each scale. It is of interest however, to
110
consider the broader themes underlying the twenty scales and
their implications for managerial development. These themes
and their implications are discussed in chapter eight.
Scale Interaction
While noting the high levels of internal consistency within
the scale categories it is important to acknowledge the
overlap that exists between them. As mentioned at the
commencement of this chapter, Pearsons Correlation
Coefficients were used to correlate each of the
questionnaire items with the initial twenty-one scale
categories. In this analysis, a number of questionnaire
items were found to load strongly and positively on more
than one of the scale categories. To further explore the
existence of interaction between the scales an additional
Pearson's Correlation analysis was conducted between the
nineteen scales (for both most and least effective
responses). The resulting correlation matrices are shown in
Appendix Four. They show clear evidence of high levels of
inter-correlation between the scales. Appendix Four
demonstrates clearly the tendency of the 170 questionnaire
items to load significantly on more than scale.
The interaction between items and scale categories,
demonstrated by the Pearson's Correlation Coefficients was
also in evidence when examining the interview data. As
111
outlined in chapters three and four, the Repertory Grid
Technique (using Hinkle's (1965) laddering approach)
facilitated the exploration of respondent construct
hierarchies. Use of the laddering questions revealed a
series of relationships between constructs in different
scale categories. Respondents frequently related managerial
strengths/weaknesses in one scale category with
strengths/weaknesses in others. No formal statistical
analysis was conducted on relationships between scales in
the interview data. However, a careful reading of the
interview transcripts provides numerous examples of scale
inter-relationships. For example, the manager's level of
technical knowledge had an impact on their confiqence and
ability to front up to decisions (i.e. they didn't have the
technical knowledge to decide quickly). Managers with low
levels of technical,knowledge were also seen as more
stressed and less able to spend time with staff. Failure to
manage stress was in turn linked with failure across a
number of other managerial dimensions. The interview data
also suggests a relationship between stress and problems
with overview, prioritisation and delegation/training.
The interview data also highlights the role of the
individual's organisational and out of work experiences, in
moulding their managerial characteristics and behaviour.
Levels of family support and related personal problems
appeared to impact on the work performance of the managers'
112
in the Department. At work, issues of being locked in to
jobs no longer enjoyed, lack of support from supervisors and
past knockbacks all contributed to lowered work capacity and
motivation. The level of support from the boss, for
example, was cited as an ingredient in managerial ability to
cope with stress. stress induced burnout in turn had an
effect on the work capacity of previously productive staff.
One respondent claimed that managers in the Social Work
Division lasted around four years before burnout became an
issue.
Both interview and questionnaire results provide clear
evidence of interaction between scale categories. This
interaction suggests that proficiency on one scale may not
only h ~ v e a short term instrumentality but may be a
prerequisite to proficiency on other scales. The patterns
of scale interaction and interdependency have not been
formally explored as part of this study. They appear to be
complex and cross all of the scale categories. The scales
in the interpersonal category seem to be of special
importance. All of the twenty categories described in
Appendix Three (with the exception of personal organisation
and technical knowledge), are arguably related to abilities
described by one or more of the interpersonal scales.
Interpersonal ability for example, underpins and serves as a
pre-requisite to effectiveness in almost all of the scales
in the conceptual ability area. Effective innovation,
113
future orientation, problem solving and overview for
example, are crucially dependent on the manager's ability to
draw on staff ideas and input, t ~ r o u g h consultation.
Interpersonal ability is, in our view, the heart and
lubricant of the interactive process. As we saw in chapter
two (process feature five) interpersonal interaction
provides an opportunity to simultaneously build and
implement managerial agendas. It is in the manager's
interaction with other people that the conceptual,
interpersonal and technical dimensions of the job come
together. High levels of interpersonal ability help the
manager to build efficiencies into a highly changeable,
fragmented and discontinuous working environment (Brewer and
Tomlinson 1964, Kotter 1982, Hales 1986, Mintzberg 1990).
If the manager is interpersonally inept such avenues of
efficiency are closed to them. Consequently, we would argue
that of all the scales described in this chapter it is those
falling in the interpersonal category that are the most
important.
The complex, interactive and interpersonal picture of
managerial effectiveness described above, conforms directly
with the interactive work process described in chapter two.
It also has profound implications for managerial
development. At one level it raises doubts about the
relevance of many of the simple two dimensional models
114
offered by management teachers (for example the contingency
model of Hersey and Blanchard, 1982). More broadly, it
provides support for those that criticise the overly
rational, analytical and simplistic thrust of much
university management education (Livingston 1971, Hayes and
Abernathy 1980, Leavitt 1983, Mintzberg 1989). These issues
are discussed further in chapter eight.
Comparing the Scale categories with Previous Research
Comparison with the content and Process Features Outlined in
Chapter Two
Chapters two and three highlighted the benefits of the
Repertory Grid Technique as a research approach. In
particular we emphasised its capacity to discover and
describe the construct systems of specific individuals and
research settings. The scale categories described in
Appendix Three reflect the advantages of the research
approach employed in this study. They comprise one of the
very few typologies descriptive of the characteristics and
behaviours of effective and ineffective managers. This
provides a sharp contrast with previous managerial research,
the great majority of which makes no attempt at all to
relate the managerial activity being described to any
measure of effectiveness. The broader literature on
managerial work, despite its limited reference to
115
effectiveness measures, provides a useful sounding board
against which the efficacy of the scale categories presented
here can be assessed. In reviewing this literature in
chapter two, we described eight content and seven process
features of managerial work. These features encompass much
of what is known about managerial work. It is of interest
that the twenty scale categories presented in this chapter,
despite being specifically descriptive of managerial
effectiveness (as opposed to managerial work) have much in
common with the features of managerial work described in
chapter two.
The issue of choice and definition of meaning in managerial
jobs (content feature two) is reflected in the overview,
goal setting, innovation,
managing/operating, prioritisation and delegation/training
scales. These categories reference different perspectives
from which the manager can view the job and different
behavioural approaches through which they can define their
activity. The technical/specialist versus generalist
manager distinction (content feature three) is reflected in
the manager/operator scale. This scale bears directly on
the establishment of an appropriate balance between
technical and managerial aspects of the job.
The informal/political nature of managerial work (content
feature five) is reflected in a number of the scale
116
categories. Political concerns emerge in the assertion,
team building and concern for others scales. For example,
the issue of fronting or not fronting up to management on
behalf of staff or clients is essentially a political
decision. The maintenance of informal, quasi-political
contacts emerges in the circulation/networking dimension of
team building and in the maintenance of peer contacts and
whole organisation focus in overview. The external
networking issue (content feature six) is reflected directly
in the external networking scale.
The innovation scale directly reflects the emerging concern
with change and leadership outlined in content feature
seven. The need for change and innovation is a key theme of
the leadership, literature. The direction setting aspects of
goal setting, future orientation, and overview and the
inspirational aspects of the team building scale all echo
dimensions of leadership defined in recent writing (Kotter
1988, 1990, Bennis 1989). The strong emphasis on
interpersonal contact and ability and the coupling of
intuition and analysis which characterises the scales in the
conceptual category are also evocative of the leadership
literature. Overall the scales developed in this study are
highly compatible with the dimensions described by recent
leadership studies. This is perhaps surprising, given the
public sector context of the study. It does however,
reflect the turbulence and change impacting on this and most
117
other New Zealand public sector organisations throughout the
1980s.
The social and affective nature of managerial work,
described in process features three and four, is directly
reflected in the scales in the interpersonal ability
category. Overall, the scales (both in themselves and in
their patterns of interaction) provide a good coverage of
key dimensions relating to interpersonal interaction and
people management and emphasize the critical importance of
interpersonal ability in management. As discussed above,
the findings of this study strongly affirm the importance of
interpersonal ability as a crucial pre-condition to
effective management.
The scales goal setting, future orientation, prioritising,
problem solving and overview touch on the intuitive skills
required to manage effectively in a complex, fragmented,
simultaneous and interactive managerial environment (see
process features five and six). As mentioned above, the
emphasis on intuition in these scales gives them a strong
affinity with the leadership literature. The high levels of
interaction between all of the scales also echoes the
interactive work environment described in process features
five and six. Aspects of the scales assertiveness,
consultation, overview and prioritisation touch on the
characteristics and behaviours needed to confront and
118
reconcile the competing demands and cross-pressures of the
managerial job (see process feature seven). Finally, as
r ~ v e a l e d in chapter six, the scales form into factor
categories which conform closely with the work of Kotter
(1982, 1988) and that of Katz (1974).
Overall, the features of managerial work detailed in chapter
two, are reflected in the characteristics and behaviours
that have emerged in this research. To this extent the
characteristics and behaviours identified in this research
are in harmony with previous research. It is also
encouraging to observe the compatibility between the
findings of this study and some of the more recent research
trends emphasizing the intuitive, inter'active and
interpersonal nature of managerial work. Despite its
quantitative emphasis we believe the study makes a
contribution to the primarily qualitative work in these
areas (see for example Bennis 1989, Mintzberg 1989, Hosking
and Fineman 1990).
comparison with Previous studies of Managerial Effectiveness
To further assess the contribution of the scale categories
presented here we can look again at the small body of
research which focuses specifically on managerial
effectiveness. Some similarity is in evidence between the
categories outlined here and those developed in earlier
119
effectiveness studies. They fit closely with the factors
developed by Roach (1956) and are compatible with the
categories d e v ~ l o p e d by Flanagan (1951) and Kay (1959). In
comparison with the categories generated in these studies
however, they have greater range and present richer
descriptions with greater detail specificity. They provide
more insight, in particular, on the nature of conceptual
ability.
Morse and Wagner (1978) present six factors which they show
to have a significant relationship with individual and
organisational effectiveness. From the limited description
provided in their paper it is difficult to accurately
compare the categories presented here with those outlined by
Morse and Wagner. All of the broad factor categories
presented by Morse and Wagner (1978) are represented in the
scale categories presented in Appendix Three. However, the
scales additionally describe a number of characteristics and
behaviours apparently not identified by Morse and Wagner,
among them personality, integrity and managing/operating.
As with the previous effectiveness research each of the
scale categories presents a richer and more detailed
description than is in evidence in the Morse and Wagner
paper.
The efficiency of the Repertory Grid approach is clearly
evidenced when compared with more recent effectiveness
120
research. Luthans et al (1985, p.259) offer very limited
descriptions of twelve managerial activities and behaviours.
These have neither the range nor .specificity of the scale
descriptions offered in Appendix Three. Only two of these
activities were found to be significantly related to
managerial success. Martinko and Gardner (1990, p.339)
provide a wide ranging classification system which includes
Mintzberg's ten roles. Only minimal descriptions are
provided of these roles and none of them were found to
relate to the performance measures employed in their study.
The number of categories that discriminate in terms of
effectiveness is clearly much lower than is the case with
this study. This provides further support for stewart and
stewart's (1981a) claim that the Repertory Grid Technique is
a highly efficient means of developing valid effectiveness
dimensions.
Overall, the scale categories outlined in Table 5:1 and
Appendix Three, demonstrate clear advantages over previous
descriptions of managerial effectiveness. They offer a
richer, more detailed description of the characteristics and
behaviours of effective and ineffective managers than has
previously been provided. Unlike previous studies, all of
the characteristic and behavioural categories discriminate
significantly between effective and ineffective managers.
Additionally, the scale categories unlike for example, many
of the leadership trait descriptions can be broken down into
121
individual items and used as performance criteria in their
own right. Consequently they are not limited to descriptive
use but have a functional utility which has been proven in
subsequent research by the author.
variations Between the Characteristics and Behaviours of
Most and Least Effective Managers
As outlined above, an additional objective of this part of
the study was to explore differences in the way respondents
perceived most and least effective in the
Department. This was approached by developing scale mean
scores and rankings for each data set and testing for
significance in mean score differences using t-tests.
Pearson's correlation were also conducted between
most and least effective ratings on each scale. We will
look at the results and implications of this analysis in
sequence.
Scale Mean Scores, Rankings and T-Tests
Scale means (ranked in order of magnitude) and standard
deviations, are presented in Table 5:2. Table 5:2 shows
that the standard deviations are substantially higher for
the least effective manager ratings than for the most
effective. Respondents range more widely over the five
point questionnaire scale when rating least effective
122
TABLE 5:2
SCALE MEAN SCORES, RANKINGS AND STANDARD DEVIATIONS
Most Effective Managers Least Effective Manaqers
Rank Scale Mean SD Rank Scale Mean SD
1. Goal Setting 4.55 .537 1. Team Building 2.40 .786
2. Assertiveness 4.49 .447 2. Innovation 2.48 .794
3. Work capacity 4.46 . 515 3 Stress Management 2.55 1. 019
4. Integrity 4.41 .478 4. Delegation/Training 2.56 .695
5. Future Orientation 4.40 .513 5. Concern for others 2.57 .825
6. Personal Organisation 4.38 . 511 6 Managing/Operating 2.61 .777
7. Technical Knowledge 4.37 .702 7. Feedback 2.62 .810
8. Team Building 4.366 .553 8. Problem Solving 2.63 .744
9. Problem Solving 4.363 ~ 5 5 3 9. Personality 2.64 .750
10. Stress Management 4.359 .687 10. Future Orientation 2.65 .843
11. Prioritising 4.358 .491 11. Goal Setting 2.66 .930
12. Overview 4.34 .513 12. Overview 2.67 .741
13. Concern for others 4.33 493 13 Consultation 2.68 .813
14. Consultation 4.31 .496 14. Prioritising 2.71 .730
15. Feedback 4.309 .559 15. Assertiveness 2.76 .785
16. Personality 4.30 .516 16. Technical Knowledge 2.82 1.111
17. Delegation/Training 4.29 .515 17. Integrity 2.85 .856
18. Innovation 4.27 .542 18. Personal Organisation 2.93 .852
19. Managing/Operating 4.15 .591 19. Work Capacity 3.00 .779
managers than when rating most effective managers. There
are a couple of possible explanations for this result. The
first explanation is that the least effective m a n a ~ e r s
varied in character and effectiveness to a greater extent
than did the most effective manager sample. The global
rating responses (recorded at the end of the questionnaire)
indicate that this is not the case. The most effective
ratees actually have a greater spread in their global
ratings (306 out of 365 of the most effective ratees receive
global ratings of four and five) than do the least effective
ratees (338 out of 365 of the least effective ratees receive
global ratings of one and two). Simply stated, the least
effective ratees are a more uniform group (in terms of
perceived effectiveness) than the most effective ratees.
The most plausible explanation for the higher variances is
.
that it is possible to be an ineffective manager in a
greater variety of ways than are available to those who want
to be effective managers. Most effective managers appear to
have a tighter and more homogeneous range of distinctive
characteristics and behaviours and hence a narrower and more
defined path of travel than do least effective managers.
This is an interesting finding as it indicates a qualitative
difference in the nature of most and least effective
management.
Mean scores for most effective managers (shown on Table 5:2)
are ranked from highest to lowest. The means for the least
123
effective managers are ranked in the opposite direction,
from lowest to highest means. The higher the score (and
ranking) the more proficient the group, is perceived to be
on a given scale. Lower scores (and rankings) indicate a
lower level of perceived proficiency on a given scale. In
both cases it is assumed that the more extreme the scale
mean the greater the significance of that scale in
describing either most or least effective managers. For
example, team building, (with a least effective manager
rank of one and a scale value of 2.4) is assumed to be of
greater significance in describing least effective managers,
than it is in describing. most effective managers, where it
has a rank of eight. Conversely goal setting is seen as
highly significant in describing most effective managers,
but of lesser significance (rank eleven) in describing least
effective managers.
A scan of Table 5:2 indicates clear differences in the
patterns of scale emphasis between most and least effective
manager ratings. In almost all cases, the ranks of the
scale categories differ between most and least effective
manager ratings. As discussed above, t-tests were used to
explore points of significant difference in the scale mean
scores and rankings. The t-test results are presented using
Duncan's (1955) New Multiple Range format in Tables 5:3 and
5:4. Tables 5:3 and 5:4 indicate that the between scale
differences are highly significant. Some broad themes can
124
be distinguished through close examination of the two
tables.
A look at the top five rated categories on Table 5:3
indicates a high level of perceived proficiency (for the
most effective manager group) on scales falling in the
conceptual category. Four of the five top ranked scales
fall into this category. Overall, it is proficiency on
scales in the conceptual category which characterise the
most effective manager group. This is reflected in an
average mean score for scales in the conceptual ability
category of 4.40 compared to 4.31 for scales in the
interpersonal ability category. While this difference is
not significant, it does highlight the stronger overall
emphasis on the conceptually orientated scales. The lower
overall mean score on scales in the interpersonal ability
category (see discussion above and in chapter six) is
reflected in the rankings shown on Table 5:3. Three of the
five lowest ranked scales are from the interpersonal ability
category. Of the bottom seven scale rankings (see Table
5:3) five fall in the interpersonal ability category. It is
in the interpersonal area that the most effective manager
group appear to be the weakest (although still significantly
ahead of the least effective manager group).
An interesting exception to these broad themes are the low
rankings of the managing-operating and innovation scales
125
'l'ULB 5:3
Or
DIFFERENCES, FOR MOST
A ax t-TESTS
(* Significance cut-off point set at the .05 level)
SCALES
V
9
1
18
I 17
16 15 14 13 12 11 10 9 S 7 6 5 4 3 2 1
MEAN .15 .27 .29 .30 .31 .31 .33 .34 .36 .36 .36 .37 .37 .3S .41 .41 .46 .49 .55
----
19. MgOp 4.15 1.2 .14 .15 .16 .16 .18 .19 .21 .21 .21 .22 .22 .23 .26 .26 .31 .34 .40
18. Innv .27 .02 .03 .04 .04 .06 .07 .09 .09 .09 .10 .10 .11 .14 .14 .19 .22 .2S
17. Deltrn .29 .01 .02 .02 .04 .05 .07 .07 .07 .OS .08 .09 .12 .12 .17 .20 .26
16. Pers .30 .01 .01 .03 .04 .06 .06 .06 .07 .07 .08 .11 .11 .16 .19 .25
15. Fdbk .31 .00 .02 .03 .05 .05 .05 .06 .06 .07 .10 .10 .15 .1S .24
14. Cons .31 .02 .03 .05 .05 .05 .06 .06 .07 .10 .10 .15 .18 .24
13. Concn .33
.01 .03 .03 .03 .04- .04 .05 .08 .08 .12 .16 .22
12. O/V .34
.02 .02 .02 .03 .03 .04 .07 .07 .07 .15 .21
11. Prior .36
.00 .00 .01 .01 .02 .05 .05 .10 .13 .19
10. Stress .36
.00 .01 .01 .02 .05 .05 .10 .13 .19
9. Prob .36
.00 .01 .02 .05 .05 .10 .13 .19
8. Team .37
.00 .01 .04 .04 .09 .12 .1S
7. TechKn .37
.01 .04 .04 .09 .12 .18
6. PrsOrg .38
.03 .03 .08 .11 .17
5. Future .41
.00 .01 .08 .14
4. Integ .41
.01 .08 .14
3. Work .46
.03 .19
2. Assert .49
.06
1. Goal 4.55
1. Goal
2. Assert
3. Work
---r--
4. Integ -----1----
5. Future
6. PersOrg
7. TechKno\ol
8. Team
9. Prob
10. Stress
11. Prior
12. O/V
13 . Concern
____ J____ ___ _
____ J____ _ ___ j ___ _
----J---- ---- ----
----
14. Consult
15. Feedback
16. Personalty
17. Deltrain
18. Innov
19. Mng/Op
====l====Ji-=::J::::1i----
I
which both fall in the conceptual category. Table 5:3
highlights these scales as the areas in which the most
effective manager group are least proficient. As can be
seen from the scale descriptions provided in Appendix Three,
both of these scales are concerned with coping with change.
The manager-operator scale is concerned with the transition
from technical specialist to broad managerial roles. The
innovation scale is concerned with coping with and
contributing to broader organisational change. Given the
very substantial changes impacting on the Department at the
time of the study it is not surprising that such change
management issues should emerge as the area of greatest
difficulty for the most effective manager group.
Turning to the least effective managers, the five lowest
ranked scales on Table 5:4 (team-building, innovation,
stress management, delegation-training and concern for
others) are indicative of a generalized difficulty with
people management, on the part of the least effective
manager group. Three of the five lowest ranked scales fall
in the interpersonal ability category. The average mean
s c o ~ e of scales in the interpersonal ability category for
the least effective manager group is 2.62 compared with 2.70
for scales in the conceptual ability category. While the
differences in mean scores are not significant, they do
highlight the greater overall emphasis on scales in the
interpersonal ability category_ Overall the least effective
126
TABLE 5:4
SIGNIFICANCE OF SCALE MEAN SCORE DIFfERENCES FOR LEAST EFFECTIVE MANAGER EATINGS
SCALES
1. Team
2. Innv
1
MEAN 12.4
2.40
.48
3. Stress .55
4. Deltrn .56
5. Concn .57
6. MngOp .61
7. Fdbk .62
8. Prob .63
9. Pers .64
10. FUture .65
11. Goal .66
12. O/V .67.
13. Cons .68
14. Prior .71
15. Assert .76
16. TechKn .82
17. Integ .88
18. PrsOrg .93
19. Work 3.00
AS ESTABLISHED BY t-TESTS
cut-off point set at .05 level)
2 i J 1 4 i 5 i 6 7
8 9 10 11 12 13 14 15
:::_ .62
.63 .64 .65 .66 .67 .68 .71 .76
.08 .15 .16 .17 .21 .22 .23 .24 .25 .26 .27 .28 .31 .36
.07 .08 .09 .13 .14 .15 .16 .:17 .18 .19 .20 .23 .28
.01 .02 .06 .07 .08 .09 .10 .12 .13 .14 .15 .16
.01 .05 .06 .07 .08 .09 .10 .12 .13 .15 .20
.04 .05 .06 .07 .08 .09 .10 .12 .14 .19
.01 .02 .03 .04 .05 .06 .07 .10 .15
.01 .02 .03 .04 .05 .06 .09 .14
.01 .02 .03 .04 .05 .08 .13
.01 .02 .03 .04 .07 .12
.01 .02 .03 .06 .11
.01 .02 .05 .10
.01 .04 .09
.03 .08
.05
,--------------------------------------------------------------------------
19. Work
18. Persorg
17. Integ
16. Tech Know
15. Assert
14. Prior
13. Consult
12. O/V
11. Goal
10. Future
9. Personal it
8. Prob
7. Feedback
6. MngjOp
5. Concern
4. Deltrain
J. Stress
2. Innov
1. Team
-